predicTTE: An accessible and optimal tool for time-to-event prediction in neurological diseases

Marcel Weinreich; Harry McDonough; Nancy Yacovzada; Iddo Magen; Yahel Cohen; Calum Harvey; Sarah Gornall; Sarah Boddy; James Alix; Nima Mohseni; Julian M Kurz; Kevin P Kenna; Sai Zhang; Alfredo Iacoangeli; Ahmad Al-Khleifat; Michael P Snyder; Esther Hobson; Ammar Al-Chalabi; Eran Hornstein; Eran Elhaik; Pamela J Shaw; Christopher McDermott; Johnathan Cooper-Knock

doi:10.1101/2024.07.20.604416

predicTTE: An accessible and optimal tool for time-to-event prediction in neurological diseases

Marcel Weinreich, Harry McDonough, Nancy Yacovzada, Iddo Magen, Yahel Cohen, Calum Harvey, Sarah Gornall, Sarah Boddy, James Alix, Nima Mohseni, Julian M Kurz, Kevin P Kenna, Sai Zhang, Alfredo Iacoangeli, Ahmad Al-Khleifat, Michael P Snyder, Esther Hobson, Ammar Al-Chalabi, Eran Hornstein, Eran ElhaikPamela J Shaw, Christopher McDermott, Johnathan Cooper-Knock

Research output: Working paper/Preprint › Preprint

Abstract

Time-to-event prediction is a key task for biological discovery, experimental medicine, and clinical care. This is particularly true for neurological diseases where development of reliable biomarkers is often limited by difficulty visualising and sampling relevant cell and molecular pathobiology. To date, much work has relied on Cox regression because of ease-of-use, despite evidence that this model includes incorrect assumptions. We have implemented a set of deep learning and spline models for time-to-event modelling within a fully customizable 'app' and accompanying online portal, both of which can be used for any time-to-event analysis in any disease by a non-expert user. Our online portal includes capacity for end-users including patients, Neurology clinicians, and researchers, to access and perform predictions using a trained model, and to contribute new data for model improvement, all within a data-secure environment. We demonstrate a pipeline for use of our app with three use-cases including imputation of missing data, hyperparameter tuning, model training and independent validation. We show that predictions are optimal for use in downstream applications such as genetic discovery, biomarker interpretation, and personalised choice of medication. We demonstrate the efficiency of an ensemble configuration, including focused training of a deep learning model. We have optimised a pipeline for imputation of missing data in combination with time-to-event prediction models. Overall, we provide a powerful and accessible tool to develop, access and share time-to-event prediction models; all software and tutorials are available at www.predictte.org.

Original language	English
DOIs	https://doi.org/10.1101/2024.07.20.604416
Publication status	Published - 23 Jul 2024

Publication series

Name	bioRxiv : the preprint server for biology
ISSN (Print)	2692-8205

Access to Document

10.1101/2024.07.20.604416

Cite this

Weinreich, M., McDonough, H., Yacovzada, N., Magen, I., Cohen, Y., Harvey, C., Gornall, S., Boddy, S., Alix, J., Mohseni, N., Kurz, J. M., Kenna, K. P., Zhang, S., Iacoangeli, A., Al-Khleifat, A., Snyder, M. P., Hobson, E., Al-Chalabi, A., Hornstein, E., ... Cooper-Knock, J. (2024). predicTTE: An accessible and optimal tool for time-to-event prediction in neurological diseases. (bioRxiv : the preprint server for biology). https://doi.org/10.1101/2024.07.20.604416

@techreport{0510ac52807f430ba0edb1261dfcbe7b,

title = "predicTTE: An accessible and optimal tool for time-to-event prediction in neurological diseases",

abstract = "Time-to-event prediction is a key task for biological discovery, experimental medicine, and clinical care. This is particularly true for neurological diseases where development of reliable biomarkers is often limited by difficulty visualising and sampling relevant cell and molecular pathobiology. To date, much work has relied on Cox regression because of ease-of-use, despite evidence that this model includes incorrect assumptions. We have implemented a set of deep learning and spline models for time-to-event modelling within a fully customizable 'app' and accompanying online portal, both of which can be used for any time-to-event analysis in any disease by a non-expert user. Our online portal includes capacity for end-users including patients, Neurology clinicians, and researchers, to access and perform predictions using a trained model, and to contribute new data for model improvement, all within a data-secure environment. We demonstrate a pipeline for use of our app with three use-cases including imputation of missing data, hyperparameter tuning, model training and independent validation. We show that predictions are optimal for use in downstream applications such as genetic discovery, biomarker interpretation, and personalised choice of medication. We demonstrate the efficiency of an ensemble configuration, including focused training of a deep learning model. We have optimised a pipeline for imputation of missing data in combination with time-to-event prediction models. Overall, we provide a powerful and accessible tool to develop, access and share time-to-event prediction models; all software and tutorials are available at www.predictte.org.",

author = "Marcel Weinreich and Harry McDonough and Nancy Yacovzada and Iddo Magen and Yahel Cohen and Calum Harvey and Sarah Gornall and Sarah Boddy and James Alix and Nima Mohseni and Kurz, {Julian M} and Kenna, {Kevin P} and Sai Zhang and Alfredo Iacoangeli and Ahmad Al-Khleifat and Snyder, {Michael P} and Esther Hobson and Ammar Al-Chalabi and Eran Hornstein and Eran Elhaik and Shaw, {Pamela J} and Christopher McDermott and Johnathan Cooper-Knock",

year = "2024",

month = jul,

day = "23",

doi = "10.1101/2024.07.20.604416",

language = "English",

series = "bioRxiv : the preprint server for biology",

type = "WorkingPaper",

}

Weinreich, M, McDonough, H, Yacovzada, N, Magen, I, Cohen, Y, Harvey, C, Gornall, S, Boddy, S, Alix, J, Mohseni, N, Kurz, JM, Kenna, KP, Zhang, S, Iacoangeli, A , Al-Khleifat, A, Snyder, MP, Hobson, E, Al-Chalabi, A, Hornstein, E, Elhaik, E, Shaw, PJ, McDermott, C & Cooper-Knock, J 2024 'predicTTE: An accessible and optimal tool for time-to-event prediction in neurological diseases' bioRxiv : the preprint server for biology. https://doi.org/10.1101/2024.07.20.604416

TY - UNPB

T1 - predicTTE

T2 - An accessible and optimal tool for time-to-event prediction in neurological diseases

AU - Weinreich, Marcel

AU - McDonough, Harry

AU - Yacovzada, Nancy

AU - Magen, Iddo

AU - Cohen, Yahel

AU - Harvey, Calum

AU - Gornall, Sarah

AU - Boddy, Sarah

AU - Alix, James

AU - Mohseni, Nima

AU - Kurz, Julian M

AU - Kenna, Kevin P

AU - Zhang, Sai

AU - Iacoangeli, Alfredo

AU - Al-Khleifat, Ahmad

AU - Snyder, Michael P

AU - Hobson, Esther

AU - Al-Chalabi, Ammar

AU - Hornstein, Eran

AU - Elhaik, Eran

AU - Shaw, Pamela J

AU - McDermott, Christopher

AU - Cooper-Knock, Johnathan

PY - 2024/7/23

Y1 - 2024/7/23

N2 - Time-to-event prediction is a key task for biological discovery, experimental medicine, and clinical care. This is particularly true for neurological diseases where development of reliable biomarkers is often limited by difficulty visualising and sampling relevant cell and molecular pathobiology. To date, much work has relied on Cox regression because of ease-of-use, despite evidence that this model includes incorrect assumptions. We have implemented a set of deep learning and spline models for time-to-event modelling within a fully customizable 'app' and accompanying online portal, both of which can be used for any time-to-event analysis in any disease by a non-expert user. Our online portal includes capacity for end-users including patients, Neurology clinicians, and researchers, to access and perform predictions using a trained model, and to contribute new data for model improvement, all within a data-secure environment. We demonstrate a pipeline for use of our app with three use-cases including imputation of missing data, hyperparameter tuning, model training and independent validation. We show that predictions are optimal for use in downstream applications such as genetic discovery, biomarker interpretation, and personalised choice of medication. We demonstrate the efficiency of an ensemble configuration, including focused training of a deep learning model. We have optimised a pipeline for imputation of missing data in combination with time-to-event prediction models. Overall, we provide a powerful and accessible tool to develop, access and share time-to-event prediction models; all software and tutorials are available at www.predictte.org.

AB - Time-to-event prediction is a key task for biological discovery, experimental medicine, and clinical care. This is particularly true for neurological diseases where development of reliable biomarkers is often limited by difficulty visualising and sampling relevant cell and molecular pathobiology. To date, much work has relied on Cox regression because of ease-of-use, despite evidence that this model includes incorrect assumptions. We have implemented a set of deep learning and spline models for time-to-event modelling within a fully customizable 'app' and accompanying online portal, both of which can be used for any time-to-event analysis in any disease by a non-expert user. Our online portal includes capacity for end-users including patients, Neurology clinicians, and researchers, to access and perform predictions using a trained model, and to contribute new data for model improvement, all within a data-secure environment. We demonstrate a pipeline for use of our app with three use-cases including imputation of missing data, hyperparameter tuning, model training and independent validation. We show that predictions are optimal for use in downstream applications such as genetic discovery, biomarker interpretation, and personalised choice of medication. We demonstrate the efficiency of an ensemble configuration, including focused training of a deep learning model. We have optimised a pipeline for imputation of missing data in combination with time-to-event prediction models. Overall, we provide a powerful and accessible tool to develop, access and share time-to-event prediction models; all software and tutorials are available at www.predictte.org.

U2 - 10.1101/2024.07.20.604416

DO - 10.1101/2024.07.20.604416

M3 - Preprint

C2 - 39091819

T3 - bioRxiv : the preprint server for biology

BT - predicTTE

ER -

predicTTE: An accessible and optimal tool for time-to-event prediction in neurological diseases

Abstract

Publication series

Access to Document

Fingerprint

Cite this