You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
abstract = {Background. Machine learning (ML), one aspect of artificial intelligence (AI), involves computer algorithms that train themselves. They have been widely applied in the healthcare domain. However, many trained ML algorithms operate as black boxes, producing a prediction from input data without a clear explanation of their workings. Non-transparent predictions are of limited utility in many clinical domains, where decisions must be justifiable. Methods. Here, we apply class-contrastive counterfactual reasoning to ML to demonstrate how specific changes in inputs lead to different predictions of mortality in people with severe mental illness (SMI), a major public health challenge. We produce predictions accompanied by visual and textual explanations as to how the prediction would have differed given specific changes to the input. We apply it to routinely collected data from a mental health secondary care provider in patients with schizophrenia. Using a data structuring framework informed by clinical knowledge, we captured information on physical health, mental health, and social predisposing factors. We then trained an ML algorithm to predict the risk of death. Results. The ML algorithm predicted mortality with an area under receiver operating characteristic curve (AUROC) of 0.8 (compared to an AUROC of 0.67 from a logistic regression model), and produced class-contrastive explanations for its predictions. Conclusions. In patients with schizophrenia, our work suggests that use of medications like second generation antipsychotics and antidepressants was associated with lower risk of death. Abuse of alcohol/drugs and a diagnosis of delirium were associated with higher risk of death. Our ML models highlight the role of co-morbidities in determining mortality in patients with SMI and the need to manage them. We hope that some of these bio-social factors can be targeted therapeutically by either patient-level or service-level interventions. This approach combines clinical knowledge, health data, and statistical learning, to make predictions interpretable to clinicians using class-contrastive reasoning. This is a step towards interpretable AI in the management of patients with SMI and potentially other diseases. {\#}{\#}{\#} Competing Interest Statement RNC consults for Campden Instruments Ltd and receives royalties from Cambridge University Press, Cambridge Enterprise, and Routledge. SB, PL and PJ declare they have no conflicts of interest to disclose. {\#}{\#}{\#} Funding Statement This work was funded by an MRC Mental Health Data Pathfinder grant (MC PC 17213). PBJ is supported by the NIHR Applied Research Collaboration East of England. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. This research was supported in part by the NIHR Cambridge Biomedical Research Centre. The views expressed are those of the authors and not necessarily those of the MRC, the NHS, the NIHR, or the Department of Health and Social Care. {\#}{\#}{\#} Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: The CPFT Research Database operates under UK NHS Research Ethics approvals (REC references 12/EE/0407, 17/EE/0442; IRAS project ID 237953). All necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines and uploaded the relevant EQUATOR Network research reporting checklist(s) and other pertinent material as supplementary files, if applicable. Yes This study reports on human clinical data which cannot be published directly due to reasonable privacy concerns, as per NHS research ethics approvals and information governance rules.},
735
+
author = {Banerjee, Soumya and Lio, Pietro and Jones, Peter B and Cardinal, Rudolf Nicholas},
736
+
doi = {10.1101/2021.04.05.21254684},
737
+
journal = {medRxiv},
738
+
mendeley-groups = {cam{\_}project,My{\_}PAPERS},
739
+
month = {apr},
740
+
pages = {2021.04.05.21254684},
741
+
publisher = {Cold Spring Harbor Laboratory Press},
742
+
title = {{A human-interpretable machine learning approach to predict mortality in severe mental illness}},
abstract = {Machine learning (ML), one aspect of artificial intelligence (AI), involves computer algorithms that train themselves. They have been widely applied in the healthcare domain. However, many trained ML algorithms operate as ‘black boxes', producing a prediction from input data without a clear explanation of their workings. Non-transparent predictions are of limited utility in many clinical domains, where decisions must be justifiable. Here, we apply class-contrastive counterfactual reasoning to ML to demonstrate how specific changes in inputs lead to different predictions of mortality in people with severe mental illness (SMI), a major public health challenge. We produce predictions accompanied by visual and textual explanations as to how the prediction would have differed given specific changes to the input. We apply it to routinely collected data from a mental health secondary care provider in patients with schizophrenia. Using a data structuring framework informed by clinical knowledge, we captured information on physical health, mental health, and social predisposing factors. We then trained an ML algorithm and other statistical learning techniques to predict the risk of death. The ML algorithm predicted mortality with an area under receiver operating characteristic curve (AUROC) of 0.80 (95{\%} confidence intervals [0.78, 0.82]). We used class-contrastive analysis to produce explanations for the model predictions. We outline the scenarios in which class-contrastive analysis is likely to be successful in producing explanations for model predictions. Our aim is not to advocate for a particular model but show an application of the class-contrastive analysis technique to electronic healthcare record data for a disease of public health significance. In patients with schizophrenia, our work suggests that use or prescription of medications like antidepressants was associated with lower risk of death. Abuse of alcohol/drugs and a diagnosis of delirium were associated with higher risk of death. Our ML models highlight the role of co-morbidities in determining mortality in patients with schizophrenia and the need to manage co-morbidities in these patients. We hope that some of these bio-social factors can be targeted therapeutically by either patient-level or service-level interventions. Our approach combines clinical knowledge, health data, and statistical learning, to make predictions interpretable to clinicians using class-contrastive reasoning. This is a step towards interpretable AI in the management of patients with schizophrenia and potentially other diseases.},
808
+
author = {Banerjee, Soumya and Lio, Pietro and Jones, Peter B. and Cardinal, Rudolf N.},
809
+
doi = {10.1038/s41537-021-00191-y},
810
+
issn = {2334-265X},
811
+
journal = {npj Schizophr.},
812
+
keywords = {Biomarkers,Schizophrenia},
813
+
mendeley-groups = {cam{\_}project,My{\_}PAPERS},
814
+
month = {dec},
815
+
number = {1},
816
+
pages = {1--13},
817
+
publisher = {Nature Publishing Group},
818
+
title = {{A class-contrastive human-interpretable machine learning approach to predict mortality in severe mental illness}},
819
+
volume = {7},
820
+
year = {2021}
821
+
}
822
+
823
+
824
+
@article{Banerjee2022,
825
+
abstract = {Achieving sufficient statistical power in a survival analysis usually requires large amounts of data from different sites. Sensitivity of individual-level data, ethical and practical considerations regarding data sharing across institutions could be a potential challenge for achieving this added power. Hence we implemented a federated meta-analysis approach of survival models in DataSHIELD, where only anonymous aggregated data are shared across institutions, while simultaneously allowing for exploratory, interactive modelling. In this case, meta-analysis techniques to combine analysis results from each site are a solution, but a manual analysis workflow hinders exploration. Thus, the aim is to provide a framework for performing meta-analysis of Cox regression models across institutions without manual analysis steps for the data providers. We introduce a package (dsSurvival) which allows privacy preserving meta-analysis of survival models, including the calculation of hazard ratios. Our tool can be of great use in biomedical research where there is a need for building survival models and there are privacy concerns about sharing data. A tutorial in bookdown format with code, diagnostics, plots and synthetic data is available here: https://neelsoumya.github.io/dsSurvivalbookdown/ All code is available from the following repositories: https://github.com/neelsoumya/dsSurvivalClient/ https://github.com/neelsoumya/dsSurvival/ {\#}{\#}{\#} Competing Interest Statement The authors have declared no competing interest.},
826
+
author = {Banerjee, Soumya and Sofack, Ghislain and Papakonstantinou, Thodoris and Avraam, Demetris and Burton, Paul and Z{\"{o}}ller, Daniela and Bishop, Tom RP},
827
+
doi = {10.1101/2022.01.04.471418},
828
+
journal = {bioRxiv},
829
+
mendeley-groups = {cam{\_}project,My{\_}PAPERS},
830
+
month = {jan},
831
+
pages = {2022.01.04.471418},
832
+
publisher = {Cold Spring Harbor Laboratory},
833
+
title = {{dsSurvival: Privacy preserving survival models for federated individual patient meta-analysis in DataSHIELD}},
834
+
year = {2022}
835
+
}
836
+
837
+
@article{Banerjee2022a,
838
+
author = {Banerjee, Soumya and Sofack, Ghislain N. and Papakonstantinou, Thodoris and Avraam, Demetris and Burton, Paul and Z{\"{o}}ller, Daniela and Bishop, Tom R. P.},
839
+
doi = {10.1186/s13104-022-06085-1},
840
+
issn = {1756-0500},
841
+
journal = {BMC Res. Notes},
842
+
mendeley-groups = {cam{\_}project,My{\_}PAPERS},
843
+
month = {dec},
844
+
number = {1},
845
+
pages = {197},
846
+
title = {{dsSurvival: Privacy preserving survival models for federated individual patient meta-analysis in DataSHIELD}},
0 commit comments