TY - JOUR
T1 - Application of SHAP values for inferring the optimal functional form of covariates in pharmacokinetic modeling
AU - for the OPTI-CLOT Study Group and SYMPHONY Consortium
AU - Janssen, A.
AU - Hoogendoorn, M.
AU - Cnossen, M.H.
AU - Mathôt, R.A.A.
AU - Cnossen, M.H.
AU - Reitsma, S.H.
AU - Leebeek, F.W.G.
AU - Mathôt, R.A.A.
AU - Fijnvandraat, K.
AU - Coppens, M.
AU - Meijer, K.
AU - Schols, S.E.M.
AU - Eikenboom, H.C.J.
AU - Schutgens, R.E.G.
AU - Beckers, E.A.M.
AU - Ypma, P.
AU - Kruip, M.J.H.A.
AU - Polinder, S.
AU - Tamminga, R.Y.J.
AU - Brons, P.
AU - Fischer, K.
AU - van Galen, K.P.M.
AU - Heubel-Moenen, F.C.J.I.
AU - Nieuwenhuizen, L.
AU - Driessens, M.H.E.
AU - van Vliet, I.
AU - Lock, J.
AU - Hazendonk, H.C.A.M.
AU - van Moort, I.
AU - Heijdra, J.M.
AU - Goedhart, M.H.J.
AU - Al Arashi, W.
AU - Preijers, T.
AU - de Jager, N.C.B.
AU - Bukkems, L.H.
AU - Cloesmeijer, M.E.
AU - Janssen, A.
AU - Collins, P.W.
AU - Liesner, R.
AU - Chowdary, P.
AU - Millar, C.M.
AU - Hart, D.
AU - Keeling, D.
AU - OPTI-CLOT study group and SYMPHONY consortium
N1 - © 2022 The Authors. CPT: Pharmacometrics & Systems Pharmacology published by Wiley Periodicals LLC on behalf of American Society for Clinical Pharmacology and Therapeutics.
PY - 2022/8
Y1 - 2022/8
N2 - In population pharmacokinetic (PK) models, interindividual variability is explained by implementation of covariates in the model. The widely used forward stepwise selection method is sensitive to bias, which may lead to an incorrect inclusion of covariates. Alternatives, such as the full fixed effects model, reduce this bias but are dependent on the chosen implementation of each covariate. As the correct functional forms are unknown, this may still lead to an inaccurate selection of covariates. Machine learning (ML) techniques can potentially be used to learn the optimal functional forms for implementing covariates directly from data. A recent study suggested that using ML resulted in an improved selection of influential covariates. However, how do we select the appropriate functional form for including these covariates? In this work, we use SHapley Additive exPlanations (SHAP) to infer the relationship between covariates and PK parameters from ML models. As a case-study, we use data from 119 patients with hemophilia A receiving clotting factor VIII concentrate peri-operatively. We fit both a random forest and a XGBoost model to predict empirical Bayes estimated clearance and central volume from a base nonlinear mixed effects model. Next, we show that SHAP reveals covariate relationships which match previous findings. In addition, we can reveal subtle effects arising from combinations of covariates difficult to obtain using other methods of covariate analysis. We conclude that the proposed method can be used to extend ML-based covariate selection, and holds potential as a complete full model alternative to classical covariate analyses.
AB - In population pharmacokinetic (PK) models, interindividual variability is explained by implementation of covariates in the model. The widely used forward stepwise selection method is sensitive to bias, which may lead to an incorrect inclusion of covariates. Alternatives, such as the full fixed effects model, reduce this bias but are dependent on the chosen implementation of each covariate. As the correct functional forms are unknown, this may still lead to an inaccurate selection of covariates. Machine learning (ML) techniques can potentially be used to learn the optimal functional forms for implementing covariates directly from data. A recent study suggested that using ML resulted in an improved selection of influential covariates. However, how do we select the appropriate functional form for including these covariates? In this work, we use SHapley Additive exPlanations (SHAP) to infer the relationship between covariates and PK parameters from ML models. As a case-study, we use data from 119 patients with hemophilia A receiving clotting factor VIII concentrate peri-operatively. We fit both a random forest and a XGBoost model to predict empirical Bayes estimated clearance and central volume from a base nonlinear mixed effects model. Next, we show that SHAP reveals covariate relationships which match previous findings. In addition, we can reveal subtle effects arising from combinations of covariates difficult to obtain using other methods of covariate analysis. We conclude that the proposed method can be used to extend ML-based covariate selection, and holds potential as a complete full model alternative to classical covariate analyses.
UR - http://www.scopus.com/inward/record.url?scp=85132689440&partnerID=8YFLogxK
U2 - https://doi.org/10.1002/psp4.12828
DO - https://doi.org/10.1002/psp4.12828
M3 - Article
SN - 2163-8306
VL - 11
SP - 1100
EP - 1110
JO - CPT: Pharmacometrics and Systems Pharmacology
JF - CPT: Pharmacometrics and Systems Pharmacology
IS - 8
ER -