Journal article 26 views
Continuous Well-being Monitoring through Interpretable Machine Learning: An Empirically Validated Framework for Inclusive Digital Governance
Frederic Boy 
,
Annie Tubadji
,
Annie Tubadji
Oxford Intersections: AI in Society, Volume: Majority World Section (editors: Rachel Adams, Fola Adeleke, Urvashi Aneja, Leah Junck)
Swansea University Authors:
Frederic Boy , Annie Tubadji
Abstract
Traditional well-being surveys reach policymakers months too late to guide effectiveintervention, leaving governments reactive rather than responsive to evolving citizenneeds. This challenge is particularly acute in the Global South, where overlappingcrises demand agile policy responses, but monitor...
| Published in: | Oxford Intersections: AI in Society |
|---|---|
| Published: |
Oxford, UK and New York, US
Oxford University Press
|
| URI: | https://cronfa.swan.ac.uk/Record/cronfa71956 |
| first_indexed |
2026-05-19T21:33:56Z |
|---|---|
| last_indexed |
2026-05-22T20:28:51Z |
| id |
cronfa71956 |
| recordtype |
SURis |
| fullrecord |
<?xml version="1.0"?><rfc1807 xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"><datestamp>2026-05-20T15:18:29.8059189</datestamp><bib-version>v2</bib-version><id>71956</id><entry>2026-05-19</entry><title>Continuous Well-being Monitoring through Interpretable Machine Learning: An Empirically Validated Framework for Inclusive Digital Governance</title><swanseaauthors><author><sid>43e704698d5dbbac3734b7cd0fef60aa</sid><ORCID>0000-0003-1373-6634</ORCID><firstname>Frederic</firstname><surname>Boy</surname><name>Frederic Boy</name><active>true</active><ethesisStudent>false</ethesisStudent></author><author><sid>f17b08e9124965486f3b5885a87b396d</sid><ORCID>0000-0002-6134-3520</ORCID><firstname>Annie</firstname><surname>Tubadji</surname><name>Annie Tubadji</name><active>true</active><ethesisStudent>false</ethesisStudent></author></swanseaauthors><date>2026-05-19</date><deptcode>CBAE</deptcode><abstract>Traditional well-being surveys reach policymakers months too late to guide effectiveintervention, leaving governments reactive rather than responsive to evolving citizenneeds. This challenge is particularly acute in the Global South, where overlappingcrises demand agile policy responses, but monitoring infrastructure remains limited.This study demonstrates that machine learning can transform readily available digitalsignals into reliable forecasts of population life satisfaction, offering a pathway tocontinuous well-being monitoring where it is most urgently needed. Validated usingUK pandemic-era data, our framework shows that both mood indicators and searchbehaviours can accurately predict life satisfaction, achieving R² = 0.72 (survey-based) and R² = 0.623 (digital behavioural), with RMSE values of 0.100 and 0.127respectively, and NRMSE of 10.4% and 13.0%. Here, ‘continuous’ denotes weeklypopulation-level monitoring, ‘interpretable’ reflects the use of an explainable LSBoostmodel providing feature-importance outputs, and ‘empirical’ indicates validation onreal-world, nationally representative survey and digital behavioural data. Digitalapproaches offering promise for resource-constrained settings. The system isexplicitly designed for adaptation across diverse cultural contexts, addressing criticalchallenges of digital equity, cultural specificity, and algorithmic sovereignty whilemaintaining transparency and local control. By enabling continuous monitoring at lowcost, this work empowers governments and civil society organizations to transitionfrom crisis management to anticipatory governance—responding to citizen well-being as it evolves rather than after damage is done.</abstract><type>Journal Article</type><journal>Oxford Intersections: AI in Society</journal><volume>Majority World Section (editors: Rachel Adams, Fola Adeleke, Urvashi Aneja, Leah Junck)</volume><journalNumber/><paginationStart/><paginationEnd/><publisher>Oxford University Press</publisher><placeOfPublication>Oxford, UK and New York, US</placeOfPublication><isbnPrint/><isbnElectronic/><issnPrint/><issnElectronic/><keywords/><publishedDay>0</publishedDay><publishedMonth>0</publishedMonth><publishedYear>0</publishedYear><publishedDate>0001-01-01</publishedDate><doi/><url/><notes/><college>COLLEGE NANME</college><department>Management School</department><CollegeCode>COLLEGE CODE</CollegeCode><DepartmentCode>CBAE</DepartmentCode><institution>Swansea University</institution><apcterm>Not Required</apcterm><funders/><projectreference/><lastEdited>2026-05-20T15:18:29.8059189</lastEdited><Created>2026-05-19T22:20:31.2411357</Created><path><level id="1">Faculty of Humanities and Social Sciences</level><level id="2">School of Management - Business Management</level></path><authors><author><firstname>Frederic</firstname><surname>Boy</surname><orcid>0000-0003-1373-6634</orcid><order>1</order></author><author><firstname>Annie</firstname><surname>Tubadji</surname><orcid>0000-0002-6134-3520</orcid><order>2</order></author></authors><documents/><OutputDurs><OutputDur><Id>359</Id><IsDataAvailableOnline xsi:nil="true"/><DataNotAvailableOnlineReasonId xsi:nil="true"/><IsDurRestrictions xsi:nil="true"/><DurRestrictionReasonId xsi:nil="true"/><DurEmbargoDate xsi:nil="true"/></OutputDur></OutputDurs></rfc1807> |
| spelling |
2026-05-20T15:18:29.8059189 v2 71956 2026-05-19 Continuous Well-being Monitoring through Interpretable Machine Learning: An Empirically Validated Framework for Inclusive Digital Governance 43e704698d5dbbac3734b7cd0fef60aa 0000-0003-1373-6634 Frederic Boy Frederic Boy true false f17b08e9124965486f3b5885a87b396d 0000-0002-6134-3520 Annie Tubadji Annie Tubadji true false 2026-05-19 CBAE Traditional well-being surveys reach policymakers months too late to guide effectiveintervention, leaving governments reactive rather than responsive to evolving citizenneeds. This challenge is particularly acute in the Global South, where overlappingcrises demand agile policy responses, but monitoring infrastructure remains limited.This study demonstrates that machine learning can transform readily available digitalsignals into reliable forecasts of population life satisfaction, offering a pathway tocontinuous well-being monitoring where it is most urgently needed. Validated usingUK pandemic-era data, our framework shows that both mood indicators and searchbehaviours can accurately predict life satisfaction, achieving R² = 0.72 (survey-based) and R² = 0.623 (digital behavioural), with RMSE values of 0.100 and 0.127respectively, and NRMSE of 10.4% and 13.0%. Here, ‘continuous’ denotes weeklypopulation-level monitoring, ‘interpretable’ reflects the use of an explainable LSBoostmodel providing feature-importance outputs, and ‘empirical’ indicates validation onreal-world, nationally representative survey and digital behavioural data. Digitalapproaches offering promise for resource-constrained settings. The system isexplicitly designed for adaptation across diverse cultural contexts, addressing criticalchallenges of digital equity, cultural specificity, and algorithmic sovereignty whilemaintaining transparency and local control. By enabling continuous monitoring at lowcost, this work empowers governments and civil society organizations to transitionfrom crisis management to anticipatory governance—responding to citizen well-being as it evolves rather than after damage is done. Journal Article Oxford Intersections: AI in Society Majority World Section (editors: Rachel Adams, Fola Adeleke, Urvashi Aneja, Leah Junck) Oxford University Press Oxford, UK and New York, US 0 0 0 0001-01-01 COLLEGE NANME Management School COLLEGE CODE CBAE Swansea University Not Required 2026-05-20T15:18:29.8059189 2026-05-19T22:20:31.2411357 Faculty of Humanities and Social Sciences School of Management - Business Management Frederic Boy 0000-0003-1373-6634 1 Annie Tubadji 0000-0002-6134-3520 2 359 |
| title |
Continuous Well-being Monitoring through Interpretable Machine Learning: An Empirically Validated Framework for Inclusive Digital Governance |
| spellingShingle |
Continuous Well-being Monitoring through Interpretable Machine Learning: An Empirically Validated Framework for Inclusive Digital Governance Frederic Boy Annie Tubadji |
| title_short |
Continuous Well-being Monitoring through Interpretable Machine Learning: An Empirically Validated Framework for Inclusive Digital Governance |
| title_full |
Continuous Well-being Monitoring through Interpretable Machine Learning: An Empirically Validated Framework for Inclusive Digital Governance |
| title_fullStr |
Continuous Well-being Monitoring through Interpretable Machine Learning: An Empirically Validated Framework for Inclusive Digital Governance |
| title_full_unstemmed |
Continuous Well-being Monitoring through Interpretable Machine Learning: An Empirically Validated Framework for Inclusive Digital Governance |
| title_sort |
Continuous Well-being Monitoring through Interpretable Machine Learning: An Empirically Validated Framework for Inclusive Digital Governance |
| author_id_str_mv |
43e704698d5dbbac3734b7cd0fef60aa f17b08e9124965486f3b5885a87b396d |
| author_id_fullname_str_mv |
43e704698d5dbbac3734b7cd0fef60aa_***_Frederic Boy f17b08e9124965486f3b5885a87b396d_***_Annie Tubadji |
| author |
Frederic Boy Annie Tubadji |
| author2 |
Frederic Boy Annie Tubadji |
| format |
Journal article |
| container_title |
Oxford Intersections: AI in Society |
| container_volume |
Majority World Section (editors: Rachel Adams, Fola Adeleke, Urvashi Aneja, Leah Junck) |
| institution |
Swansea University |
| publisher |
Oxford University Press |
| college_str |
Faculty of Humanities and Social Sciences |
| hierarchytype |
|
| hierarchy_top_id |
facultyofhumanitiesandsocialsciences |
| hierarchy_top_title |
Faculty of Humanities and Social Sciences |
| hierarchy_parent_id |
facultyofhumanitiesandsocialsciences |
| hierarchy_parent_title |
Faculty of Humanities and Social Sciences |
| department_str |
School of Management - Business Management{{{_:::_}}}Faculty of Humanities and Social Sciences{{{_:::_}}}School of Management - Business Management |
| document_store_str |
0 |
| active_str |
0 |
| description |
Traditional well-being surveys reach policymakers months too late to guide effectiveintervention, leaving governments reactive rather than responsive to evolving citizenneeds. This challenge is particularly acute in the Global South, where overlappingcrises demand agile policy responses, but monitoring infrastructure remains limited.This study demonstrates that machine learning can transform readily available digitalsignals into reliable forecasts of population life satisfaction, offering a pathway tocontinuous well-being monitoring where it is most urgently needed. Validated usingUK pandemic-era data, our framework shows that both mood indicators and searchbehaviours can accurately predict life satisfaction, achieving R² = 0.72 (survey-based) and R² = 0.623 (digital behavioural), with RMSE values of 0.100 and 0.127respectively, and NRMSE of 10.4% and 13.0%. Here, ‘continuous’ denotes weeklypopulation-level monitoring, ‘interpretable’ reflects the use of an explainable LSBoostmodel providing feature-importance outputs, and ‘empirical’ indicates validation onreal-world, nationally representative survey and digital behavioural data. Digitalapproaches offering promise for resource-constrained settings. The system isexplicitly designed for adaptation across diverse cultural contexts, addressing criticalchallenges of digital equity, cultural specificity, and algorithmic sovereignty whilemaintaining transparency and local control. By enabling continuous monitoring at lowcost, this work empowers governments and civil society organizations to transitionfrom crisis management to anticipatory governance—responding to citizen well-being as it evolves rather than after damage is done. |
| published_date |
0001-01-01T17:20:53Z |
| _version_ |
1866631011286122496 |
| score |
11.106612 |