Hostname: page-component-586b7cd67f-gb8f7 Total loading time: 0 Render date: 2024-11-23T17:26:45.646Z Has data issue: false hasContentIssue false

Interdisciplinary methods for researching climate-mental health links through the Methane Early Warning Network (ME-NET): improving ‘visibility’ and integrating complex multi-datasets

Published online by Cambridge University Press:  25 October 2024

A response to the following question: What are the likely impacts of climate change on rates of depression and other mood disorders? What actions can be taken by individuals, communities or nations to reduce those impacts?

Harriet Elizabeth Moore*
Affiliation:
Lincoln Institute for Rural and Coastal Health, University of Lincoln, College of Health and Science, Lincoln, UK Development, Inequalities, Resilience, and Environments (DIRE) Research Group, Lincoln, UK
John Atanbori
Affiliation:
School of Engineering and Physical Sciences, University of Lincoln, College of Health and Science, Lincoln, UK
Ebenezer Forkuo Amankwaa
Affiliation:
Department of Geography and Resource Development, University of Ghana, Legon, Greater Accra, Ghana
Mark Gussy
Affiliation:
Lincoln Institute for Rural and Coastal Health, University of Lincoln, College of Health and Science, Lincoln, UK Development, Inequalities, Resilience, and Environments (DIRE) Research Group, Lincoln, UK
Aloysius Niroshan Siriwardena
Affiliation:
School of Health and Social Care, University of Lincoln, College of Health and Science, Lincoln, UK Community and Health Research Unit (CaHRU), Lincoln, UK
Edward Hanna
Affiliation:
School of Natural Science, University of Lincoln, College of Health and Science, Lincoln, UK Lincoln Climate Research Group, Lincoln, UK
*
Corresponding author: Harriet Elizabeth Moore; Email: [email protected]
Rights & Permissions [Opens in a new window]

Abstract

Climatic and atmospheric conditions impact mental health, including increased incidents of depression associated with air pollution. A growing body of research considers time-bound ‘snap-shots’ of climatic drivers and mental health outcomes. Less is known about the likely effects of future climate change on mental health. Research is often inhibited by data scarcity, the challenge of synthesising data across multiple geospatial and temporal scales, and the under-representation of hard-to-reach groups. Thus, research methods are needed to integrate and analyse complex environmental and mental health multi-datasets while improving the visibility of under-represented groups. In this methods paper we present a novel approach for investigating the impact of climate change on mental health and addressing some challenges with, a) invisibility of vulnerable groups, and b) integrating complex environmental and mental health multi-datasets. The research aim is to pilot a web-based and smartphone application (Methane Early Warning Network (ME-NET)) for investigating the role of methane as a precursor of on-ground ozone, and the impact of ozone on mental health outcomes to improve civic knowledge and health-protection behaviour in the United Kingdom and Ghana. The methods include exploring the feasibility of using machine learning to develop an ozone early warning system and application co-design.

Type
Results
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s) 2024. Published by Cambridge University Press

Introduction

Climatic and atmospheric conditions impact mental health both directly, such as the effect of heat on severe depression, suicide (Thompson et al., Reference Thompson, Hornigold, Page and Waite2018) and psychosis (Hansen et al., Reference Hansen, Bi, Nitschke, Ryan, Pisaniello and Tucker2008), and indirectly via interactions between meteorological factors and human activity, such as the development of polluting ozone in the air people breath (known as ‘ground level ozone’). Polluting emissions like methane, nitrous oxides and other non-volatile organic compounds combine with high temperatures to produce ozone (Francoeur et al., Reference Francoeur, McDonald, Gilman, Zarzana, Dix, Brown, de Gouw, Frost, Li, McKeen and Peischl2021) which impacts mental health outcomes (Bernardini et al., Reference Bernardini, Attademo, Trezzi, Gobbicchi, Balducci, Del Bello, Menculini, Pauselli, Piselli, Sciarma and Moretti2020). Ground level concentrations of ozone are expected to increase under climate change warming scenarios (Hertig, Reference Hertig2020). Studies suggest that ozone exposure is associated with increased psychiatric symptoms (Qiu et al., Reference Qiu, Danesh-Yazdi, Weisskopf, Kosheleva, Spiro, Wang, Coull, Koutrakis and Schwartz2022) and admissions (Bernardini et al., Reference Bernardini, Attademo, Trezzi, Gobbicchi, Balducci, Del Bello, Menculini, Pauselli, Piselli, Sciarma and Moretti2020), and rates of mood disorders like depression (Zhao et al., Reference Zhao, Tesch, Markevych, Baumbach, Janssen, Schmitt, Romanos, Nowak and Heinrich2020). However, the underlying biological mechanisms for how ozone impacts mental health are poorly understood (Nguyen et al., Reference Nguyen, Malig and Basu2021), and much of the research in this area is inconclusive (Zhao et al., Reference Zhao, Markevych, Romanos, Nowak and Heinrich2018). The study reported in this manuscript addresses the research question, ‘What are the likely impacts of climate change on rates of depression and other mood disorders?’ by exploring links between ground level ozone and mental health outcomes.

In the context of climate change, mental health conditions are expected to account for an increasing proportion of global health burden in coming decades (Vigo et al., Reference Vigo, Thornicroft and Atun2016; Wu et al., Reference Wu, Wang, Tao, Cao, Yuan, Ye, Chen, Wang and Zhu2023). While the field of climate and health is advancing rapidly, compared to physical health outcomes like respiratory disease, less research has explored links between climate change and mental health outcomes (Massazza et al., Reference Massazza, Teyton, Charlson, Benmarhnia and Augustinavicius2022). Of research that does examine mental health outcomes, most studies focus on the impact of temperature, involve ‘snapshots’ of climatic condition rather than the longer-term process of climate change explicitly and consider environmental exposures at low geographical resolution. Conducting higher resolution research is often inhibited by data scarcity of one or more important variables (e.g., Kolstad & Johansson, Reference Kolstad and Johansson2011), and the challenge of synthesising data across multiple geospatial and temporal scales (Daraio & Glänzel, Reference Daraio and Glänzel2016; Massazza et al., Reference Massazza, Teyton, Charlson, Benmarhnia and Augustinavicius2022).

Further challenges relate to the under-representation of hard-to-reach groups in mental health research (Chamberlain & Hodgetts, Reference Chamberlain and Hodgetts2022; Dowrick et al., Reference Dowrick, Gask, Edwards, Aseem, Bower, Burroughs, Catlin, Chew-Graham, Clarke, Gabbay and Gowers2009). As a result, the mental health challenges linked to climate and experienced by most vulnerable people and regions are often ‘invisible’. Thus, innovative research methods are urgently needed to illuminate the effects of climate change currently and in the coming decades on mental health, particularly for those individuals, communities and regions that are already at greatest risk from environmental and meteorological exposures. Understanding impacts is a first step to developing innovative climate change mitigation and adaptation interventions with mental and physical health co-benefits, including improving public health preparedness to reduce climate-related disaster risk (Hess et al., Reference Hess, Errett, McGregor, Busch Isaksen, Wettstein, Wheat and Ebi2023). In this methods paper, we present a novel methodological approach for conducting research about the impact of climate change on mental health that addresses some challenges associated with, a) invisibility and the heterogeneity of mental health impacts between and within groups and individuals most vulnerable to climate change, and b) integrating multiple complex environmental and mental health datasets to form ‘multi-datasets’.

Challenge One: Invisibility and heterogeneity

Traditional recruitment strategies present challenges for researching the mental health outcomes of climate change because they tend to result in the under-representation or exclusion of most vulnerable people (Borderon et al., Reference Borderon, Best, Bailey, Hopping, Dove and Cervantes de Blois2021). Climate change vulnerability and socio-economic exclusion often co-occur (Sevoyan & Hugo, Reference Sevoyan and Hugo2014; Resurrección et al., Reference Resurrección, Goodrich, Song, Bastola, Prakash, Joshi, Liebrand and Shah2019; Ifeanyi-Obi & Ugorji, Reference Ifeanyi-Obi and Ugorji2020). Most studies of symptoms and conditions related to depression and mental health outcomes more generally are conducted with participants characterised as ‘Western, Educated, Industrialised, Rich, and Democratic’, or WEIRD (Henrich et al., Reference Henrich, Heine and Norenzayan2010; Apicella et al., Reference Apicella, Norenzayan and Henrich2020). Thus, most data and knowledge about the drivers of severe outcomes represent regions with the greatest capacity for climate adaptation like the G7 countries (Masuda et al., Reference Masuda, Batdorj and Senzaki2020), while ‘hard-to-reach’ people and regions, including low-to-middle-income regions (LMIRs) (Masuda et al., Reference Masuda, Batdorj and Senzaki2020) as well as marginalised groups in high-income regions (Apicella et al., Reference Apicella, Norenzayan and Henrich2020), are excluded from evolving narratives and evidence bases. These ‘invisible’ people and places include individuals and groups in geographic regions that are likely to experience the most severe and immediate impacts of climate change. For example, increasingly prolonged periods of hot-humid days associated with climate change are expected in dry climatic zones, such as sub-Saharan Africa (SSA), while record-breaking heatwaves are anticipated in Afghanistan and regions of Australia including remote rural areas (Thompson et al., Reference Thompson, Mitchell, Hegerl, Collins, Leach and Slingo2023).

Compared to more economically developed places like Australia with centralised healthcare, research about the impact of climate and climate change on mental health in LMIRs like SSA is extremely limited. This is partly because health services in SSA are comparatively decentralised and privatised, and service use data often grossly underrepresent mental health condition rates in the region; The Lancet Commission on the Future of Health in SSA (Agyepong et al., Reference Agyepong, Sewankambo, Binagwaho, Coll-Seck, Corrah, Ezeh, Fekadu, Kilonzo, Lamptey, Masiye and Mayosi2017) estimates that the treatment gap may be as high as 90% for severe mental health conditions. As a result, many of the retrospective observational data science approaches that are common in regions with centralised health care systems like Australia and the UK are less viable in SSA and other LMIRs with colonial legacies and fragmented health care systems. Thus, research methods are needed that are robust to variations in health services and address persistent inequalities between research ecosystems in high-income compared to LMIRs, as well as the exclusion of ‘hard-to-reach’ communities in more affluent regions. These methodological developments should take place in conjunction with the wider goal of the global research community to decolonise data science (Khan, Reference Khan2022).

In addition to differences between high-income regions and LMIRs, the impacts of climate and climate change on acute mental illnesses like major depressive disorder are also likely to be heterogenous within regions, such as between and within rural and urban areas (Oğur, Reference Oğur2023), demographically between people of varying age, ethnicity and gender (Liu et al., Reference Liu, Varghese, Hansen, Xiang, Zhang, Dear, Gourley, Driscoll, Morgan, Capon and Bi2021), socio-economically between wealthier and more deprived communities (Sapari et al., Reference Sapari, Selamat, Isa, Ismail and Mahiyuddin2023), and related to complex individual differences that make up the life course histories of people, including the presence or absence of coping mechanisms (Bonanno et al., Reference Bonanno, Brewin, Kaniasty and Greca2010). Further, it is probable that the experience of climatic conditions like severe heat varies between regions due to subjectivity. Differences in subjective heat stress occur depending on prior encounters with temperature such that continued exposure to high temperatures produces an acclimation effect (Daanen et al., Reference Daanen, Racinais and Périard2018). People and communities in regions characterised by high temperatures may experience less heat stress associated with climate change compared to those in regions where high temperatures are rare. Similarly, temperatures that are considered ‘extreme’ vary regionally (Liu et al., Reference Liu, Varghese, Hansen, Xiang, Zhang, Dear, Gourley, Driscoll, Morgan, Capon and Bi2021), partly in relation to ordinary regional temperature ranges. For example, heat warnings are issued in England for forecasting two days of maximum 30°C and minimum 15°C, and in the Netherlands for more than a 10% probability of four or more days with a maximum exceeding 27°C (Casanueva et al., Reference Casanueva, Burgstall, Kotlarski, Messeri, Morabito, Flouris, Nybo, Spirig and Schwierz2019). By comparison, in central and southern Greece, ‘yellow’, ‘amber’ and ‘red’ heat warnings are issued for forecasts of maximum 37°C, 41°C, greater than 44°C respectively. Variation is also likely exacerbated by the urban heat island effect, with 56% of people now living in cities (World Bank, 2023), such that temperatures in homes and hospitals may vary considerably compared regional averages (Gough et al., Reference Gough, Yankson and Wilby2019). Thus, the temperatures that precipitate the escalation of acute psychological conditions may vary between people, regions and small areas, although multi-region studies of subjective heat stress are limited. The heterogeneity of lived experience and of mental health outcomes associated with climatic conditions reinforces the need for research methods that are robust to both regional differences, and differences within demographic, socio-economic, and geographic groups.

Challenge Two: analysing complex multi-datasets

Identifying individuals, groups and regions that are likely to be most at risk of experiencing severe mental health impacts because of climate change, at scale, will require moving beyond traditional methods of conducting mental health research, including common study designs. Most studies involve single-region (Rocque et al., Reference Rocque, Beaudoin, Ndjaboue, Cameron, Poirier-Bergeron, Poulin-Rheault, Fallon, Tricco and Witteman2021) or single-sample qualitative methods, longitudinal designs, case-control, single-sample repeated measures, or cross-sectional analysis (Burrows et al., Reference Burrows, Denckla, Hahn, Schiff, Okuzono, Randriamady, Mita, Kubzansky, Koenen and Lowe2024; Charlson et al., Reference Charlson, Ali, Benmarhnia, Pearl, Massazza, Augustinavicius and Scott2021). These approaches provide invaluable geographic or time-bound ‘snapshots’ about the determinants of mental health outcomes. However, determinants linked to climatic events and climate change are occurring over unprecedented spatial and temporal scales, necessitating some additional methodological approaches to capture multi-scalar variation (Massazza et al., Reference Massazza, Teyton, Charlson, Benmarhnia and Augustinavicius2022). Further, data granularity, quality and completeness often vary between environmental and health datasets (Cui et al., Reference Cui, Eccles, Kwok, Joubert, Messier and Balshaw2022) within single regions as well as between regions. Thus, in addition to improving visibility, novel methods are needed for integrating and analysing complex multi-datasets that capture the temporal and spatial scale of climate-driven processes with mental health outcomes.

Advanced computational methods like machine learning (ML) and deep learning (DL) are well suited to multi-data analysis, particularly for datasets with challenging characteristics, such as multi-scalar variation. To date, most climate and health research utilising advanced computational methods focus on physical health outcomes like infectious disease (e.g., Boudreault et al., Reference Boudreault, Campagna and Chebana2023; Schneider et al., Reference Schneider, Sebastianelli, Spiller, Wheeler, Carmo, Nowakowski, Garcia-Herranz, Kim, Barlevi, El Raiss Cordero and Liberata Ullo2021) rather than mental health. Of those studies that do consider mental health outcomes, most analyse global (e.g., Pizzulli et al., Reference Pizzulli, Telesca and Covatariu2021) or single region (Fahim et al., Reference Fahim, Uddin, Ahmed, Islam and Ahmed2022) aggregate data rather than multi-regional or small area data. Machine learning techniques offer several key benefits in addressing challenges associated with multi-regions, multi-datasets and variation in data quality and completeness. They excel in uncovering complex patterns and relationships within large and diverse datasets (e.g., Du et al., Reference Du, Li, Yang and Horng2019). These techniques provide predictive capabilities for various outcomes, enhancing our understanding of the interplay between health and climate factors. Moreover, ML can be utilised for the automatic interpolation of data, helping to mitigate issues arising from data scarcity, as highlighted by Li et al. (Reference Li, Zhang, Holt, Tian and Piltner2011). Additionally, these techniques automate the process of feature extraction and selection, aiding in the identification of important features for analysis (Qi et al., Reference Qi, Wang, Song, Hu, Li and Zhang2018).

The transdisciplinary research methodology presented here was developed to deliver the project ‘Methane Early Warning Network’ (ME-NET) funded by the Wellcome Trust (228267/Z/23/Z) to improve understanding of the effect of methane on health outcomes, including mental health presentations, in the UK and Ghana. The study sites capture the varying opportunities and obstacles associated with multi-data research in both high-income regions and LMIRs. The approach addresses the challenges described above in two ways. Firstly, representation of ordinarily ‘invisible’ and underrepresented communities and individuals will be achieved by adopting participatory and ethnographic methods (Graham et al., Reference Graham, Kane and Gussy2022), including ME-NET application co-design with lived experience experts and community champions. Engaging lived experience experts and community champions addresses the limitations of relying exclusively on service use data in regions like Ghana where mental health condition rates are underrepresented in service use datasets.

The research team undertook conceptual development through regular consultation with known stakeholders about project parameters (e.g., exploring regional preferences for web-based compared to phone-based applications and establishing health and environmental data availability), including the Ghana Meteorological Agency, Ghana Health Services and NHS representatives in the UK. The health and environmental sector stakeholders, in addition to representatives of coastal and regional communities in Lincolnshire and Ghana will attend stakeholder engagement boards in both regions for the iterative development and co-design of application functions, ensuring cultural relevance, usability and accessibility for diverse groups and individuals within and between regions. Community representatives will also support the research group to identify target users, including members of health support groups and community centres.

Secondly, the research integrates geospatial data visualisation methods (Moore et al., Reference Moore, Hill, Siriwardena, Law, Thomas, Gussy, Spaight and Tanser2022a, Reference Khan2022b) and ML approaches to elucidate relationships between ‘messy’ health datasets and open environmental datasets while maintaining data integrity. Finally, the interactive application will be developed equally for web-based and mobile phone apps to facilitate accessibility across diverse regions. Smart-phone penetration is higher in some LMIRs like Ghana (140% penetration in 2022 meaning nearly 50% of the population operates multiple mobile phones (FurtherAfrica, 2022)) compared to high-income regions like the UK (984% penetration in 2020 (ONS, 2022)). In many LMIRs, smartphones are used in lieu of other devices like laptops and home computers. Populations routinely utilise smartphone applications for diverse purposes, including receiving environmental warnings (e.g., flood alerts), and for health monitoring (e.g., mHEALTH (Blay et al., Reference Blay, Amankwaa, Afolabi and Mensah2023)). The prevalence of smartphone use in LMIRs can facilitate the participation of communities and regions who have traditionally been excluded from global data science conversations, thus overcoming many of the limitations of traditional ‘WEIRD’ recruitment. Unlike many contemporary health applications, all ME-NET functions will be equally accessible via a web platform and smartphone to ensure accessibility for digitally excluded groups like ageing coastal communities in the UK who are less likely to utilise phone applications.

Methods

Research aims and questions

The aim of the research is to pilot a web-based and smartphone application (ME-NET) for regions with varying environmental and mental health data availability and quality, and with varying sources of methane emitters for. The prototype application will be tested for up to twelve months, including monthly evaluation in-built evaluation modules enabling iterative live functionality development. The purpose of the application is two-fold, a) developing data synthesis approaches for understanding the impact of climate and climate change on mental health outcomes that are globally applicable, b) training methane ‘early warning’ models for improving civic knowledge and health-protection behaviour that are robust to regional contexts. The study addresses three research questions:

  1. 1. To what extent can ML, including DL be used to develop an ozone early warning system that incorporates mental health data in two regions of the world with a) higher (UK), and b) lower-to-middle income (Ghana), reflecting wider global variation in data availability and quality?

  2. 2. Given available mental health data in the UK and Ghana, is it viable to use DL to predict rates of mental health emergencies associated with air quality, and if so, what impact do methane and ozone have on severe mental health presentations?

  3. 3. How do lived experiences of how mental health symptoms associated with methane and ozone vary geospatially?

Location

The regional focus is communities in two areas in Lincolnshire, UK and two areas in Ghana, Africa. Two regions were chosen to facilitate app development for data rich and data scarce contexts. The UK produces some of the most temporally and geographically granular and high-quality open health data in the world, reflecting research ecosystems in high-income regions. Open health data collated in Ghana is of varying quality and granularity, reflecting LMIR research ecosystems. Ghana also has one of the highest smartphone penetration rates in the world. Multiple areas per region have been selected to, a) ensure PPIE and LEE groups capture regional within and between group diversity, and b) to explore direct and distal relationships between environmental conditions and health outcomes. Grimsby on the Lincolnshire coast is located near waste disposal sites, coal mining and oil processing, while the City of Lincoln is located inland, south of Grimsby’s industrial operations. Sekondi-Takoradi (ST) and Accra are located on Ghana coastline. Industrial operations (e.g., oil and gas processing) occur near ST while Accra is located further East and is the capital of Ghana. Thus, the research involves areas with similar distal relationships between methane production and populations.

Population

The first population involved in the research is patients with healthcare records related to mental health, excluding those who have opted out of data sharing for research purposes (e.g., UK NHS patients). Patient records include hospital attendances in Accra and ST, Ghana, and ambulance and hospital attendances in Lincolnshire, UK. The second population involved in the research includes participants in stakeholder engagement boards in the UK and Ghana, such as representatives from meteorological agencies, health services and community groups known to the research team. The third population involved in the research includes anonymous users of the application in both regions.

Recruitment

Recruitment relates to the second and third population described above and will be conducted in two phases. The first phase is purposive and involves inviting known stakeholders with lived experience of policy and practice in health, environment and climate sectors, as well as community lived experience experts to attend stakeholder engagement boards on behalf of the wider social and professional groups they represent. This phase of recruitment has been completed, drawing on existing networks and partnerships within and beyond the research team. The second phase of recruitment is opportunistic and involves ‘snowballing’, inviting members of wider social and professional groups to test and disseminate the ‘ME-NET’ application prototype. Members of the stakeholder engagement board will share the prototype via QR code with professional and social networks, including carer support groups, community wellbeing services and health centres. This phase of recruitment will be undertaken in parallel to the project co-design process, such as encouraging stakeholders involved in board meetings to identify potential user groups and distribute QR codes to group leaders and representatives.

Application co-development and functionality

The application will be co-designed and co-developed with lived experience experts in the UK and Ghana to ensure usability, accessibility and cultural relevance of individual functions to improve the uptake, and therefore visibility, of communities that might otherwise be excluded from mental health and climate change narratives. Co-design will follow the principles of ‘design justice’ and community participation (Costanza-Chock, Reference Costanza-Chock2024), including digital development through iterative processes of building, testing and adjusting application functions (Common Knowledge, 2024). Application content and functions will be tested with stakeholders during engagement board meetings. Input facilitating iterative adjustment will be collated through live polls and surveys during and following meetings. While some core application functions will be consistent between regional interfaces, the design and parameters of specific features will vary in order to maximise cultural relevance.

Similar to the techniques employed in our previous works (Aliyu et al., Reference Aliyu, Choudhury, Sohani, Atanbori, Ribeiro, Ahmed and Mishra2023; Atanbori & Rose, Reference Atanbori and Rose2022), this research will pioneer the use of ML, encompassing the exploration of artificial neural networks, deep and incremental learning, on climate data captured by satellites. This analysis will be complemented by incorporating lived experiences to train an ‘early warning’ model that integrates multi-region and multi-datasets, encompassing the best available current, historical and future mental health data. The ML model will be integrated into a developed prototype smartphone and web-based application to improve understanding of associations between methane and ozone concentrations and mental health outcomes, educate application users about links between climate and mental health, predict on-ground concentrations of ozone, and communicate health warnings to users. Educational modules, including interactive mapping functions displaying aggregate historic mental health and climate data, will be delivered through the ‘Explore and Learn’ dashboard to improve climate and health literacy among users. The dashboard will display an interactive world map with functions enabling users to select spatial layers visualising rates of mental health conditions together with concentrations of atmospheric chemicals and conditions linked to poor health outcomes, including methane, nitrous oxide and UV levels.

Incremental Learning will be used to incorporate incoming self-reported anonymised user lived experiences to improve accuracy and identify environmental thresholds linked to mental health outcomes. Individual users can register with ‘My profile’ to access a range of service options, including opting in for generic daily or weekly updates and health recommendations, and inputting regular self-reported health experiences for tracking the relationship between individual health outcomes and environmental conditions via the ‘My Health Today’ function. An advanced option, ‘Train ME-NET’, will allow users to identify environmental thresholds related to poor health outcomes and set threshold-based alerts for early warning purposes.

Data and measures

Datasets and sources for both study regions are those obtained from the Sentinel 5P satellite mission, including meteorological and atmospheric chemical concentration data. Additional weather and health datasets vary by region. The date range of datasets utilised for ML purposes and the development of educational modules is between 2018 and 2024, although some datasets are only available from 2019 onwards. Data generated from MethaneSat following the recent satellite launch will be integrated as new data are released. The size and temporal characteristics of each dataset will vary depending on completeness and availability. All data collection, storage, use and management will comply with the University of Lincoln’s Data Protection and Information Compliance requirements, following the UK General Data Protection Regulation (UK GDPR) and the Data Protection Act 2018 (DPA 2018) and equivalent requirements for Ghana’s data regulations (e.g., Data Protection Act, 2012 (ACT 843)).

Sentinel 5P satellite data

Meteorological data will be obtained from Level 1B (L1B) and Level 2 (L2) products. Atmospheric chemical concentration data will be obtained from Level 2 (L2) products. Further details about these products can be found via Sentinel Online (https://sentinels.copernicus.eu/web/sentinel/data-products). Product summary: L1B products: Irradiance product Ultraviolet Near-Infrared (UVN) module (L1B_IR_UVN), Irradiance product Short Wave Infrared (SWIR) module (L1B_IR_SIR). L2 products: Tropospheric Ozone (L2_O3_TCL), Total column Nitrogen Dioxide (L2_NO2_), Total column Sulphur Dioxide (L2_SO2_), Total column Carbon Monoxide (L2_CO_), Total column Methane (L2_CH4_), Total column Formaldehyde (L2_HCHO_), Ultraviolet (UV) Aerosol Index (L2_AER_AI), Cloud fraction, albedo, top pressure (L2_CLOUD_), NASA Suomi-NPP Program Visible Infrared Imaging Radiometer Suite (VIIRS) Clouds (L2_NP_BDx, x = 3, 6, 72).

Observational mental health data

Mental health data from Lincolnshire will include anonymised aggregate ambulance data obtained from the East Midlands Ambulance NHS Trust (EMAS), and primary care data collated by the Lincolnshire County Council Data Hub (https://lhih.org.uk/). Measures will include records of mental health emergencies related to depression, suicidality and other presentations attended by ambulances (>100,000 records) and primary care records of GP and hospital attendances related to mental health in Lincolnshire. Mental health data from Ghana will include regional aggregate anonymised health records for conditions like depression collated by Ghana Health Services (https://www.moh.gov.gh/ghana-health-service/) including the mental health division, Ghana Mental Health Authority, and aggregate anonymised health facility data (e.g., psychiatric hospitals). Data selection will take into consideration the exclusion of ‘hidden’ and ‘hard-to-reach’ groups and individuals from routine scheduled health service datasets like GP appointments. Thus, the research will utilise emergency medical service data, including ambulance and hospital admissions data which reflects acute mental health conditions, overcoming many of the limitations of ‘opt in’ service data, such as self-selection bias which typically results in the underrepresentation of groups like men and males who are less likely to engage with help-seeking (Moore et al., Reference Moore, Siriwardena, Gussy, Tanser, Hill and Spaight2021).

Self-reported mental health data

Anonymous users of the ME-NET application in the UK and Ghana will self-report health symptoms, including symptoms related to depression and anxiety by responding to survey items prompted through the ‘My Daily Health’ platform. Measures will include items adapted from tools including the General Health Questionnaire (GHQ-12), General Anxiety Disorder-7 (GAD-7), Patient Health Questionnaire-9 (PHQ-9), and the Depression, Anxiety and Stress Scale-21 (DASS-21). The application will be accessible via website and smartphone application. This approach challenges the characterisation of LMIRs as ‘data scarce’. It has potential to improve the visibility of people and communities in LMIRs and to reposition these regions as global leaders in the use of low-cost and high-penetration innovative solutions like smartphone applications to combat health disparities and reduce health service access inequalities. Smart-phone penetration is higher in some LMIRs like Ghana (140% penetration in 2022 meaning nearly 50% of the population operates multiple mobile phones (FurtherAfrica, 2022)) compared to high-income regions like the UK (95% penetration in 2021 (ONS, 2022)). In many LMIRs, smartphones are used in lieu of other devices like laptops and home computers. Populations routinely utilise smartphone applications for diverse purposes, including receiving environmental warnings (e.g., flood alerts), and for health monitoring (e.g., mHEALTH (Blay et al., Reference Blay, Amankwaa, Afolabi and Mensah2023)). The prevalence of smartphone use in LMIRs can facilitate the participation of communities and regions who have traditionally been excluded from global data science conversations, thus overcoming many of the limitations of traditional ‘WEIRD’ recruitment. Unlike many contemporary health applications, all ME-NET functions will be equally accessible via a web platform and smartphone to ensure accessibility for digitally excluded groups like ageing coastal communities in the UK who are less likely to utilise phone applications.

Additional environmental data

Environmental data for Lincolnshire will include daily live and historic weather data collated by The Met (http://www.weatherobs.com/; https://ogimet.com/gsynres.phtml.en), and DEFRA (https://uk-air.defra.gov.uk/data/ozone-databut; https://uk-air.defra.gov.uk/data/uv-data). Environmental data from Ghana will include live and historic weather data such as temperature, clouds and wind (https://www.tide-forecast.com/; https://meteologix.com/gh/observations).

Data linkage

To compile the linked dataset at lower super output area (LSOA) scale, health and environmental data will be merged using the join tool in ArcGIS Pro 2.6.0. The join will use Lower Super Output Area codes (LSOA11CD) as these identifiers are consistent between databases used in the research.

Data analysis

Machine learning: a) to predict on-ground ozone from historic environmental data, b) for integrating health data to understand climate–health relationships, c) to determine whether predicted health outcomes reflect lived experiences of health outcomes and d) to explore likely health outcomes under climate change scenarios of 1.5° and 2° warming for the two regions involved in the research. The specific ML approach will be informed by the quality of individual datasets obtained and the synergies between individual datasets (e.g., scale, completeness) which will determine the parameters for collating and analysing multi-datasets. This final analytical approach will consider the role of UV as a precursor of on-ground ozone which is linked to mental health outcomes.

Overall, the methodology champions the de-colonisation of data science by pioneering geospatial and analytical methods that capture the heterogeneity of mental health impacts associated with climate change, are robust to regional contexts, and validates the utility of lived experience in regions where more traditional health data are ‘scarce’.

Data availability statement

Some of the data that support the findings of this study will be openly available in repositories. Other data will be unavailable in compliance with ethical requirements.

Author contributions

All authors contributed to concept development. HM, JA, MG, NS and EH contributed to methodological approach and study design. HM and JA developed initial manuscript draft. All authors edited manuscript drafts. All authors are involved in project delivery.

Financial support

This research is funded by the Wellcome Trust (228267/Z/23/Z) through the Wellcome Data Science Ideathon, Climate and Health Award.

Ethical standards

This research has ethical approval from the University of (redacted) (Ethics Approval ID: 16191).

Competing interests

The authors have no conflicts of interest to declare.

References

Connections references

Hickie, I., Ospina-Pinillos, L., Scott, E., Merikangas, K., Iorfino, F., & Crouse, J. (2023). What are the likely impacts of climate change on rates of depression and other mood disorders? What actions can be taken by individuals, communities or nations to reduce those impacts? Research Directions: Depression, 1, E3. https://doi.org/10.1017/dep.2023.25 CrossRefGoogle Scholar

References

Agyepong, IA, Sewankambo, N, Binagwaho, A, Coll-Seck, AM, Corrah, T, Ezeh, A, Fekadu, A, Kilonzo, N, Lamptey, P, Masiye, F and Mayosi, B (2017) The path to longer and healthier lives for all Africans by 2030: the Lancet Commission on the future of health in sub-Saharan Africa. The Lancet 390(10114), 28032859.CrossRefGoogle ScholarPubMed
Aliyu, AM, Choudhury, R, Sohani, B, Atanbori, J, Ribeiro, JX, Ahmed, SKB and Mishra, R (2023) An artificial neural network model for the prediction of entrained droplet fraction in annular gas-liquid two-phase flow in vertical pipes. International Journal of Multiphase Flow 164, 104452.CrossRefGoogle Scholar
Apicella, C, Norenzayan, A and Henrich, J (2020) Beyond WEIRD: A review of the last decade and a look ahead to the global laboratory of the future. Evolution and Human Behavior 41(5), 319329.CrossRefGoogle Scholar
Atanbori, J and Rose, S (2022) MergedNET: A simple approach for one-shot learning in siamese networks based on similarity layers. Neurocomputing 509, 110.CrossRefGoogle Scholar
Bernardini, F, Attademo, L, Trezzi, R, Gobbicchi, C, Balducci, PM, Del Bello, V, Menculini, G, Pauselli, L, Piselli, M, Sciarma, T and Moretti, P (2020) Air pollutants and daily number of admissions to psychiatric emergency services: evidence for detrimental mental health effects of ozone. Epidemiology and Psychiatric Sciences 29, e66.CrossRefGoogle Scholar
Blay, KB, Amankwaa, EF, Afolabi, OD and Mensah, P (2023) Managing COVID-19 and health vulnerabilities: mHealth user experience, information quality and policy recommendations. International Development Planning Review 45(3), 249272.CrossRefGoogle Scholar
Bonanno, GA, Brewin, CR, Kaniasty, K and Greca, AML (2010) Weighing the costs of disaster: Consequences, risks, and resilience in individuals, families, and communities. Psychological Science in the Public Interest 11(1), 149.CrossRefGoogle ScholarPubMed
Borderon, M, Best, KB, Bailey, K, Hopping, DL, Dove, M and Cervantes de Blois, CL (2021) The risks of invisibilization of populations and places in environment-migration research. Humanities and Social Sciences Communications 8(1). doi: 10.1057/s41599-021-00999-0.CrossRefGoogle Scholar
Boudreault, J, Campagna, C and Chebana, F (2023) Machine and deep learning for modelling heat-health relationships. Science of The Total Environment 892, 164660.CrossRefGoogle ScholarPubMed
Burrows, K, Denckla, CA, Hahn, J, Schiff, JE, Okuzono, SS, Randriamady, H, Mita, C, Kubzansky, LD, Koenen, KC and Lowe, SR (2024) A systematic review of the effects of chronic, slow-onset climate change on mental health. Nature Mental Health 2(2), 228–243.CrossRefGoogle Scholar
Casanueva, A, Burgstall, A, Kotlarski, S, Messeri, A, Morabito, M, Flouris, AD, Nybo, L, Spirig, C and Schwierz, C (2019) Overview of existing heat-health warning systems in Europe. International Journal of Environmental Research and Public Health 16(15), 2657.CrossRefGoogle ScholarPubMed
Chamberlain, K and Hodgetts, D (2022) Challenges in designing qualitative research when working with and for hard-to-reach groups. U. Flick The SAGE Handbook of Qualitative Research Design 2, 10001014.CrossRefGoogle Scholar
Charlson, F, Ali, S, Benmarhnia, T, Pearl, M, Massazza, A, Augustinavicius, J and Scott, JG (2021) Climate change and mental health: a scoping review. International Journal of Environmental Research and Public Health 18(9), 4486.CrossRefGoogle ScholarPubMed
Common Knowledge (2024) Our Values. Common Knowledge. https://commonknowledge.coop/about/#our-values, accessed 5th August, 2024.Google Scholar
Costanza-Chock, S (2024) Design Practices: ‘Nothing about us without us’, Design Justice, MIT Press. https://designjustice.mitpress.mit.edu/pub/cfohnud7/release/4, accessed 5th August, 2024.Google Scholar
Cui, Y, Eccles, KM, Kwok, RK, Joubert, BR, Messier, KP and Balshaw, DM (2022) Integrating multiscale geospatial environmental data into large population health studies: challenges and opportunities. Toxics 10(7), 403.CrossRefGoogle ScholarPubMed
Daanen, HA, Racinais, S and Périard, JD (2018) Heat acclimation decay and re-induction: a systematic review and meta-analysis. Sports Medicine 48, 409430.CrossRefGoogle ScholarPubMed
Daraio, C and Glänzel, W (2016) Grand challenges in data integration—State of the art and future perspectives: An introduction. Scientometrics 108(1), 391400.CrossRefGoogle Scholar
Dowrick, C, Gask, L, Edwards, S, Aseem, S, Bower, P, Burroughs, H, Catlin, A, Chew-Graham, C, Clarke, P, Gabbay, M and Gowers, S (2009) Researching the mental health needs of hard-to-reach groups: managing multiple sources of evidence. BMC Health Services Research 9, 112.CrossRefGoogle ScholarPubMed
Du, S, Li, T, Yang, Y and Horng, SJ (2019) Deep air quality forecasting using hybrid deep learning framework. IEEE Transactions on Knowledge and Data Engineering 33(6), 24122424.CrossRefGoogle Scholar
Fahim, M, Uddin, ME, Ahmed, R, Islam, MR and Ahmed, N (2022) A machine learning based analysis between climate change and human health: A correlational study. In 2022 International Conference on Computer and Applications (ICCA) (pp. 16). IEEE.CrossRefGoogle Scholar
Francoeur, CB, McDonald, BC, Gilman, JB, Zarzana, KJ, Dix, B, Brown, SS, de Gouw, JA, Frost, GJ, Li, M, McKeen, SA and Peischl, J (2021) Quantifying methane and ozone precursor emissions from oil and gas production regions across the contiguous US. Environmental Science & Technology 55(13), 91299139.CrossRefGoogle ScholarPubMed
FurtherAfrica (2022) African Countries with the Highest Number of Mobile Phones, Africa, Communications. https://furtherafrica.com/2022/07/19/african-countries-with-the-highest-number-of-mobile-phones/, accessed 5th April, 2024.Google Scholar
Gough, KV, Yankson, PWK, Wilby, RL, et al. (2019) Vulnerability to extreme weather events in cities: implications for infrastructure and livelihoods. Journal of the British Academy 7(s2), 155–81. doi: 10.5871/jba/007s2.155.Google Scholar
Graham, F, Kane, R, Gussy, M, et al. (2022) Recovery of health and wellbeing in rural cancer survivors following primary treatment: Analysis of UK qualitative interview data. Nursing Reports 12(3), 482497. doi: 10.3390/nursrep12030046.CrossRefGoogle ScholarPubMed
Hansen, A, Bi, P, Nitschke, M, Ryan, P, Pisaniello, D and Tucker, G (2008) The effect of heat waves on mental health in a temperate Australian city. Environmental Health Perspectives, 116(10), 13691375.CrossRefGoogle Scholar
Henrich, J, Heine, SJ and Norenzayan, A (2010) Most people are not WEIRD. Nature 466(7302), 2929.CrossRefGoogle Scholar
Hertig, E (2020) Health-relevant ground-level ozone and temperature events under future climate change using the example of Bavaria, Southern Germany. Air Quality, Atmosphere & Health 13(4), 435446.CrossRefGoogle Scholar
Hess, JJ, Errett, NA, McGregor, G, Busch Isaksen, T, Wettstein, ZS, Wheat, SK and Ebi, KL (2023) Public health preparedness for extreme heat events. Annual Review of Public Health 44, 301321.CrossRefGoogle ScholarPubMed
Ifeanyi-Obi, CC and Ugorji, EC (2020) Effect of social exclusion on climate change adaptation of female arable crop farmers in Abia State, Nigeria. South African Journal of Agricultural Extension 48(1), 5569.Google Scholar
Jackson, RB, Saunois, M, Bousquet, P, Canadell, JG, Poulter, B, Stavert, AR, Bergamaschi, P, Niwa, Y, Segers, A and Tsuruta, A (2020) Increasing anthropogenic methane emissions arise equally from agricultural and fossil fuel sources. Environmental Research Letters 15(7), 071002.CrossRefGoogle Scholar
Khan, SA (2022) Decolonising global health by decolonising academic publishing. BMJ Global Health 7(3), e007811.CrossRefGoogle ScholarPubMed
Kolstad, EW and Johansson, KA (2011) Uncertainties associated with quantifying climate change impacts on human health: a case study for diarrhea. Environmental Health Perspectives 119(3), 299305.CrossRefGoogle Scholar
Li, L, Zhang, X, Holt, JB, Tian, J and Piltner, R (2011) Spatiotemporal Interpolation Methods for Air Pollution Exposure. SARA.Google Scholar
Liu, J, Varghese, BM, Hansen, A, Xiang, J, Zhang, Y, Dear, K, Gourley, M, Driscoll, T, Morgan, G, Capon, A and Bi, P (2021) Is there an association between hot weather and poor mental health outcomes? A systematic review and meta-analysis. Environment International 153, 106533.CrossRefGoogle ScholarPubMed
Massazza, A, Teyton, A, Charlson, F, Benmarhnia, T and Augustinavicius, JL (2022) Quantitative methods for climate change and mental health research: Current trends and future directions. The Lancet Planetary Health 6(7), e613e627.CrossRefGoogle ScholarPubMed
Masuda, T, Batdorj, B and Senzaki, S (2020) Culture and attention: Future directions to expand research beyond the geographical regions of WEIRD cultures. Frontiers in Psychology 11, 490858.CrossRefGoogle ScholarPubMed
Moore, HE, Hill, B, Siriwardena, N, Law, G, Thomas, C, Gussy, M, Spaight, R and Tanser, F (2022a) An exploration of factors characterising unusual spatial clusters of COVID-19 cases in the East Midlands region, UK: A geospatial analysis of ambulance 999 data. Landscape and Urban Planning 219, 104299.CrossRefGoogle Scholar
Moore, HE, Hill, B, Tanser, F, Siriwardena, AN, Gussy, M, Cutts, M and Spaight, R (2022b) Characterizing unusual spatial clusters of male mental health emergencies occurring during the first national COVID-19 “Lockdown” in the East Midlands Region, UK: A geospatial analysis of ambulance 999 data. American Journal of Men’s Health 16(3), 15579883221097539.CrossRefGoogle Scholar
Moore, HE, Siriwardena, AN, Gussy, M and Spaight, R (2023) Mental health emergencies attended by ambulances in the United Kingdom and the implications for health service delivery: A cross-sectional study. Journal of Health Services Research & Policy 28(2), 138146.CrossRefGoogle ScholarPubMed
Moore, HE, Siriwardena, AN, Gussy, M, Tanser, F, Hill, B and Spaight, R (2021) Mental health emergencies and COVID-19: the impact of ‘lockdown’ in the East Midlands of the UK. BJPsych Open, 7(4), e139.CrossRefGoogle Scholar
Nguyen, AM, Malig, BJ and Basu, R (2021) The association between ozone and fine particles and mental health-related emergency department visits in California, 2005–2013. PloS One 16(4), e0249675.CrossRefGoogle ScholarPubMed
Oğur, AA (2023) Uneven resilience of urban and rural areas to heatwaves. Journal of Design for Resilience in Architecture and Planning 4(Special Issue), 7894.CrossRefGoogle Scholar
ONS (2022) Percentage of homes and individuals with technological equipment. Data and Analysis from Census 2021. https://www.beta.ons.gov.uk/aboutus/transparencyandgovernance/freedomofinformationfoi/percentageofhomesandindividualswithtechnologicalequipment Google Scholar
Pizzulli, VA, Telesca, V and Covatariu, G (2021) Analysis of correlation between climate change and human health based on a machine learning approach. Healthcare 9(1), 86.CrossRefGoogle ScholarPubMed
Qi, Z, Wang, T, Song, G, Hu, W, Li, X and Zhang, Z (2018) Deep air learning: Interpolation, prediction, and feature analysis of fine-grained air quality. IEEE Transactions on Knowledge and Data Engineering, 30(12), 22852297.CrossRefGoogle Scholar
Qiu, X, Danesh-Yazdi, M, Weisskopf, M, Kosheleva, A, Spiro, A, Wang, C, Coull, BA, Koutrakis, P and Schwartz, JD (2022) Associations between air pollution and psychiatric symptoms in the Normative Aging Study. Environmental Research Letters 17(3), 034004.CrossRefGoogle ScholarPubMed
Resurrección, BP, Goodrich, CG, Song, Y, Bastola, A, Prakash, A, Joshi, D, Liebrand, J and Shah, SA (2019) In the shadows of the Himalayan mountains: persistent gender and social exclusion in development. The Hindu Kush Himalaya assessment: Mountains, Climate Change, Sustainability and People, 491516.CrossRefGoogle Scholar
Rocque, RJ, Beaudoin, C, Ndjaboue, R, Cameron, L, Poirier-Bergeron, L, Poulin-Rheault, RA, Fallon, C, Tricco, AC and Witteman, HO (2021) Health effects of climate change: an overview of systematic reviews. BMJ Open, 11(6), e046333.CrossRefGoogle ScholarPubMed
Sapari, H, Selamat, MI, Isa, MR, Ismail, R and Mahiyuddin, WRW (2023) The impact of heat waves on health care services in low-or middle-income countries: protocol for a systematic review. JMIR Research Protocols 12(1), e44702.CrossRefGoogle ScholarPubMed
Schneider, R, Sebastianelli, A, Spiller, D, Wheeler, J, Carmo, R, Nowakowski, A, Garcia-Herranz, M, Kim, D, Barlevi, H, El Raiss Cordero, Z and Liberata Ullo, S (2021) Climate-based ensemble machine learning model to forecast dengue epidemics. In ICML 2021 Workshop on Tackling Climate Change with Machine Learning.Google Scholar
Sevoyan, A and Hugo, G (2014) Vulnerability to climate change among disadvantaged groups: the role of social exclusion. Applied Studies in Climate Adaptation, 258265. doi: 10.1002/9781118845028.ch29.CrossRefGoogle Scholar
Thompson, R, Hornigold, R, Page, L and Waite, T (2018) Associations between high ambient temperatures and heat waves with mental health outcomes: a systematic review. Public Health 161, 171191.CrossRefGoogle ScholarPubMed
Thompson, V, Mitchell, D, Hegerl, GC, Collins, M, Leach, NJ and Slingo, JM (2023) The most at-risk regions in the world for high-impact heatwaves. Nature Communication 14, 2152.CrossRefGoogle ScholarPubMed
Thornicroft, G (2007) Most people with mental illness are not treated. The Lancet 370(9590), 807808.CrossRefGoogle Scholar
Vigo, D, Thornicroft, G and Atun, R (2016) Estimating the true global burden of mental illness. The Lancet Psychiatry 3(2), 171178.CrossRefGoogle ScholarPubMed
World Bank (2023) Urban development. https://www.worldbank.org/en/topic/urbandevelopment/overview, accessed 13th May, 2024.Google Scholar
Wu, Y, Wang, L, Tao, M, Cao, H, Yuan, H, Ye, M, Chen, X, Wang, K and Zhu, C (2023) Changing trends in the global burden of mental disorders from 1990 to 2019 and predicted levels in 25 years. Epidemiology and Psychiatric Sciences 32, e63.CrossRefGoogle ScholarPubMed
Zhao, T, Markevych, I, Romanos, M, Nowak, D and Heinrich, J (2018) Ambient ozone exposure and mental health: a systematic review of epidemiological studies. Environmental Research 165, 459472.CrossRefGoogle ScholarPubMed
Zhao, T, Tesch, F, Markevych, I, Baumbach, C, Janssen, C, Schmitt, J, Romanos, M, Nowak, D and Heinrich, J (2020) Depression and anxiety with exposure to ozone and particulate matter: An epidemiological claims data analysis. International Journal of Hygiene and Environmental Health 228, 113562.CrossRefGoogle ScholarPubMed

Author comment: Interdisciplinary methods for researching climate-mental health links through the Methane Early Warning Network (ME-NET): improving ‘visibility’ and integrating complex multi-datasets — R0/PR1

Comments

No accompanying comment.

Review: Interdisciplinary methods for researching climate-mental health links through the Methane Early Warning Network (ME-NET): improving ‘visibility’ and integrating complex multi-datasets — R0/PR2

Comments

This paper outlines an important research project addressing several significant challenges in the climate change and mental health field. The outcomes of this research - and its methodology - will be of important value to both the climate and mental health research field and to advancing support for people experiencing the mental health impacts of climate change. The paper is very well written, clear and accessible.

My comments are minor - just flagging a few areas that I think would benefit from greater clarity:

1) I was not clear from the paper over what timescale the data is being collected (appreciating that the authors note some of this will depend on data completeness and availability). For example, over what period will users be asked to self-report health symptoms for this data to be integrated/compared? How long will the prototype app be tested for before evaluation?

2) It wasn't until the end of the paper that I felt fully clear about who the target users are of the application itself - some slight re-ordering to make this clearer earlier on in the 'summary of the research' I think would be helpful. Participation of researchers in the application is also mentioned - is the app targeted to only civic knowledge and health protection behaviours, or also to connect researchers with access to data etc?

3) The authors have given thoughtful consideration to ensuring appropriate representation of those often excluded from research - the co-design and centering of lived experience expertise and data is a real strength of the approach. Will this involve one co-design process to create a standardised application version across the UK and Ghana contexts, or will elements of design be different depending on the results of the co-design on what can maximise accessibility and cultural relevance in the two contexts?

4) The authors mention the inclusion of additional weather and health datasets - I assume this will enable the analysis to account for other known influences on mental health outcomes sucvh

Is there a proposed application of this work to influence the development of climate and mental health integrated national level health surveillance and early warning systems? Climate-informed health surveillance and early warning systems for mental and psychosocial health are fairly scarce (see the WHO Climate and Health survey 2021) and this work could be hugely beneficial in advancing knowledge of how such integrated systems can be implemented and the importance of doing so.

Presentation

Overall score 4 out of 5
Is the article written in clear and proper English? (30%)
5 out of 5
Is the data presented in the most useful manner? (40%)
4 out of 5
Does the paper cite relevant and related articles appropriately? (30%)
5 out of 5

Context

Overall score 5 out of 5
Does the title suitably represent the article? (25%)
5 out of 5
Does the abstract correctly embody the content of the article? (25%)
5 out of 5
Does the introduction give appropriate context and indicate the relevance of the results to the question or hypothesis under consideration? (25%)
5 out of 5
Is the objective of the experiment clearly defined? (25%)
5 out of 5

Results

Overall score 4 out of 5
Is sufficient detail provided to allow replication of the study? (50%)
4 out of 5
Are the limitations of the experiment as well as the contributions of the results clearly outlined? (50%)
3 out of 5

Review: Interdisciplinary methods for researching climate-mental health links through the Methane Early Warning Network (ME-NET): improving ‘visibility’ and integrating complex multi-datasets — R0/PR3

Comments

Are the findings novel?

If this is an analysis of previously analysed results, is the new contribution novel and relevant?

Are the controls used in the experiment valid?

Is the data presented in the most useful manner?

Are the limitations of the experiment as well as the contributions of the results clearly outlined?

Dear authors, this is an interesting manuscript addressing important challenges as you outlined. There are issues with the structure and further content needed in the methods; please find detailed comments below.

Abstract

- P. 2 line 8 – 9: “Little is understood about the likely effects of climate change on mental health”; there is actually large amount of research in this field now so I don’t think this statement quite rings true. You can definitely still talk about the challenges that exist in the field. Note this also applies to where you’ve stated this in the introduction.

- P. 2 line 18: define ME-NET

- P. 2 line 18 – 22: Long sentence and a bit unclear. Perhaps split up to clearly define the research aim and then the implications.

- Overall, the abstract is quite repetitive and mostly focused on context. The background information could be condensed and more information on the methods could be provided.

Introduction and methods

- You begin the introduction talking about the impact of extreme weather on mental health, however, I’m not sure how relevant this is given the main exposures in your proposed methodology are methane and ozone. I would suggest starting the introduction talking about climate change and mental health broadly, then provide context about methane and ozone – you can use some of the information you provided in the start of your Methods section under ‘Summary of the research’.

- You should also reference the Question paper that you’re addressing; as stated in the Author instructions, under introduction – “Brief background of the rationale and prior research for the study, highlights how the study addresses the specific Question paper”.

- Be more explicit that this is a methods paper (and also mention this in the abstract) e.g. p. 2 line 27, “In this Methods Paper…”

- This is a paper that’s highly relevant to this work: Massazza A, Teyton A, Charlson F, Benmarhnia T, Augustinavicius JL. Quantitative methods for climate change and mental health research: current trends and future directions. Lancet Planet Health. 2022 Jul;6(7):e613-e627. https://doi.org/10.1016/S2542-5196(22)00120-6 . I would recommend reading it if you haven’t already and using it as a reference, for example, when you describe ‘Challenge Two’ and the limitations of existing studies/methods.

- There is no need for the ‘Summary of the research’ section in the methods. The Methods section should explicitly focus on the methodology; all the context and rationale should be provided in the introduction. Think about all the details that another researcher would need in order to replicate the study.

- P. 5 line 192: you mention that respiratory data will be incorporated in the model; is this as an outcome that will be investigated? It is not mentioned anywhere else in the manuscript.

- The research aims and questions should be moved to the end of the introduction.

- P.4, paragraph beginning line 146, you mention that this methodology was developed to deliver the ME-NET project, and explain some steps that have been done including undertaking conceptual development with stakeholders. It would be good to make it clearer at what stage the project currently is, and perhaps in your methods describe the process of recruiting stakeholders and how you conducted the conceptual development. Are these stakeholders the same as the ‘stakeholder engagement

boards’ mentioned on p. 7 line 270? If so, make this clear. You’ve also stated on p. 5 line 159 that these boards will be involved in the co-design of the ME-NET application, and then on p. 6 line 212 that the application will be co-designed with lived experience experts. In your methods, you should describe this co-design process in more detail.

- P. 6 line 221: I’m not sure how smartphone penetration can be 140%; I went to check the reference and it was not listed in the References section.

- Another missing reference on p. 5 line 154 (Graham 2022); please ensure all references cited are in the References section.

- P. 7 paragraph starting line 286, ‘Sentinel 5P satellite data’. This is quite a technical list; it would be good to explain in a way that someone who’s not familiar with this type of data would understand, and there are also a number of acronyms and terms that have not been defined e.g. UVN, SWIR, L1b, L2. Rather than just a list, have a paragraph like you did for the mental health data.

- On p. 3, line 75, you stated that in SSA, “Service use data often grossly underrepresent mental health condition rates in the region”. It was interesting to note that for the observational mental health data from Ghana, you’ve used data from health services. It would be good to explain this choice in relation to this issue.

- On the use of the term ‘multi-datasets’; this is not a term that I’m familiar with and a quick google suggests it’s not so commonly used. If it is referring to a specific type of dataset, it would be good to provide a definition.

Overall, your Introduction should be merged with the ‘Summary of the research’ and ‘Research aims, questions’ from your Methods, with repetition removed. For example, you could have a paragraph or two each addressing the following: context around climate change and mental health, and methane and ozone; the research challenges (which I think can be condensed); background to your proposed methods (i.e. machine learning, smart-phone context) and how this addresses the challenges; ending with research aims and questions. Then your methods section can start of with a brief explanation of the project context (what’s been done so far), location, population, data sources and then the steps that you will take to develop the platform/application and analyse the data. Be consistent with your terminology around ME-NET, as you’ve called it an “integrated data platform” in the abstract and aims, and an application elsewhere. Also looking back at your research questions, it’s a bit unclear how you are going to address these – Q1: does this simply refer to the development of the application, or will you be assessing its effectiveness? I think Q2 and 3 relate to your data analysis, in terms of predictions and measuring impacts. Your data analysis section in the Methods needs to be elaborated on accordingly, and is currently quite vague – what does ‘for integrating health data to understand climate-health relationships’ mean exactly, and how will you determine whether predicted health outcomes reflect lived experiences of health outcomes.

Presentation

Overall score 4 out of 5
Is the article written in clear and proper English? (30%)
5 out of 5
Is the data presented in the most useful manner? (40%)
4 out of 5
Does the paper cite relevant and related articles appropriately? (30%)
5 out of 5

Context

Overall score 5 out of 5
Does the title suitably represent the article? (25%)
5 out of 5
Does the abstract correctly embody the content of the article? (25%)
5 out of 5
Does the introduction give appropriate context and indicate the relevance of the results to the question or hypothesis under consideration? (25%)
5 out of 5
Is the objective of the experiment clearly defined? (25%)
5 out of 5

Results

Overall score 4 out of 5
Is sufficient detail provided to allow replication of the study? (50%)
4 out of 5
Are the limitations of the experiment as well as the contributions of the results clearly outlined? (50%)
3 out of 5

Decision: Interdisciplinary methods for researching climate-mental health links through the Methane Early Warning Network (ME-NET): improving ‘visibility’ and integrating complex multi-datasets — R0/PR4

Comments

Please take note of the extensive commentary of reviewer 2 and in particular the note about little is known about the mental health effects of climate change. As this is a rapidly developing field, it may be more appropriate to note that there are many new results and methods appearing in the public domain currently. So please provide, a more detailed response to reviewer 2 and adjust the manuscript accordingly.

Author comment: Interdisciplinary methods for researching climate-mental health links through the Methane Early Warning Network (ME-NET): improving ‘visibility’ and integrating complex multi-datasets — R1/PR5

Comments

No accompanying comment.

Decision: Interdisciplinary methods for researching climate-mental health links through the Methane Early Warning Network (ME-NET): improving ‘visibility’ and integrating complex multi-datasets — R1/PR6