Development and Optimization of a Machine-Learning Prediction Model for Acute Desquamation After Breast Radiation Therapy in the Multicenter REQUITE Cohort

Mahmoud Aldraimli; Sarah Osman; Diana Grishchuck; Samuel Ingram; Robert Lyon; Anil Mistry; Jorge Oliveira; Robert Samuel; Leila E A Shelley; Daniele Soria; Miriam V Dwek; Miguel E Aguado-Barrera; David Azria; Jenny Chang-Claude; Alison Dunning; Alexandra Giraldo; Sheryl Green; Sara Gutiérrez-Enríquez; Carsten Herskind; Hans van Hulle; Maarten Lambrecht; Laura Lozza; Tiziana Rancati; Victoria Reyes; Barry S Rosenstein; Dirk de Ruysscher; Maria C de Santis; Petra Seibold; Elena Sperk; R Paul Symonds; Hilary Stobart; Begoña Taboada-Valadares; Christopher J Talbot; Vincent J L Vakaet; Ana Vega; Liv Veldeman; Marlon R Veldwijk; Adam Webb; Caroline Weltens; Catharine M West; Thierry J Chaussalet; Tim Rattay; REQUITE consortium

doi:10.1016/j.adro.2021.100890

Development and Optimization of a Machine-Learning Prediction Model for Acute Desquamation After Breast Radiation Therapy in the Multicenter REQUITE Cohort

Mahmoud Aldraimli, Sarah Osman, Diana Grishchuck, Samuel Ingram, Robert Lyon, Anil Mistry, Jorge Oliveira, Robert Samuel, Leila E A Shelley, Daniele Soria, Miriam V Dwek, Miguel E Aguado-Barrera, David Azria, Jenny Chang-Claude, Alison Dunning, Alexandra Giraldo, Sheryl Green, Sara Gutiérrez-Enríquez, Carsten Herskind, Hans van HulleMaarten Lambrecht, Laura Lozza, Tiziana Rancati, Victoria Reyes, Barry S Rosenstein, Dirk de Ruysscher, Maria C de Santis, Petra Seibold, Elena Sperk, R Paul Symonds, Hilary Stobart, Begoña Taboada-Valadares, Christopher J Talbot, Vincent J L Vakaet, Ana Vega, Liv Veldeman, Marlon R Veldwijk, Adam Webb, Caroline Weltens, Catharine M West, Thierry J Chaussalet, Tim Rattay^*, REQUITE consortium

^*Corresponding author for this work

Research output: Contribution to journal › Article › Academic › peer-review

Abstract

Purpose: Some patients with breast cancer treated by surgery and radiation therapy experience clinically significant toxicity, which may adversely affect cosmesis and quality of life. There is a paucity of validated clinical prediction models for radiation toxicity. We used machine learning (ML) algorithms to develop and optimise a clinical prediction model for acute breast desquamation after whole breast external beam radiation therapy in the prospective multicenter REQUITE cohort study.

Methods and Materials: Using demographic and treatment-related features (m = 122) from patients (n = 2058) at 26 centers, we trained 8 ML algorithms with 10-fold cross-validation in a 50:50 random-split data set with class stratification to predict acute breast desquamation. Based on performance in the validation data set, the logistic model tree, random forest, and naïve Bayes models were taken forward to cost-sensitive learning optimisation.

Results: One hundred and ninety-two patients experienced acute desquamation. Resampling and cost-sensitive learning optimisation facilitated an improvement in classification performance. Based on maximising sensitivity (true positives), the "hero" model was the cost-sensitive random forest algorithm with a false-negative: false-positive misclassification penalty of 90:1 containing m = 114 predictive features. Model sensitivity and specificity were 0.77 and 0.66, respectively, with an area under the curve of 0.77 in the validation cohort.

Conclusions: ML algorithms with resampling and cost-sensitive learning generated clinically valid prediction models for acute desquamation using patient demographic and treatment features. Further external validation and inclusion of genomic markers in ML prediction models are worthwhile, to identify patients at increased risk of toxicity who may benefit from supportive intervention or even a change in treatment plan.

Original language	English
Article number	100890
Number of pages	13
Journal	Advances in Radiation Oncology
Volume	7
Issue number	3
DOIs	https://doi.org/10.1016/j.adro.2021.100890
Publication status	Published - 2 Jun 2022

Keywords

ACUTE SKIN TOXICITY
BIOMARKERS
NORMAL TISSUE
RADIOTHERAPY
RISK-FACTOR

Access to Document

10.1016/j.adro.2021.100890Licence: CC BY

Cite this

Aldraimli, M., Osman, S., Grishchuck, D., Ingram, S., Lyon, R., Mistry, A., Oliveira, J., Samuel, R., Shelley, L. E. A., Soria, D., Dwek, M. V., Aguado-Barrera, M. E., Azria, D., Chang-Claude, J., Dunning, A., Giraldo, A., Green, S., Gutiérrez-Enríquez, S., Herskind, C., ... REQUITE consortium (2022). Development and Optimization of a Machine-Learning Prediction Model for Acute Desquamation After Breast Radiation Therapy in the Multicenter REQUITE Cohort. Advances in Radiation Oncology, 7(3), Article 100890. https://doi.org/10.1016/j.adro.2021.100890

@article{e286424a81b64de79d21440bc2ad029e,

title = "Development and Optimization of a Machine-Learning Prediction Model for Acute Desquamation After Breast Radiation Therapy in the Multicenter REQUITE Cohort",

abstract = "Purpose: Some patients with breast cancer treated by surgery and radiation therapy experience clinically significant toxicity, which may adversely affect cosmesis and quality of life. There is a paucity of validated clinical prediction models for radiation toxicity. We used machine learning (ML) algorithms to develop and optimise a clinical prediction model for acute breast desquamation after whole breast external beam radiation therapy in the prospective multicenter REQUITE cohort study.Methods and Materials: Using demographic and treatment-related features (m = 122) from patients (n = 2058) at 26 centers, we trained 8 ML algorithms with 10-fold cross-validation in a 50:50 random-split data set with class stratification to predict acute breast desquamation. Based on performance in the validation data set, the logistic model tree, random forest, and na{\"i}ve Bayes models were taken forward to cost-sensitive learning optimisation.Results: One hundred and ninety-two patients experienced acute desquamation. Resampling and cost-sensitive learning optimisation facilitated an improvement in classification performance. Based on maximising sensitivity (true positives), the {"}hero{"} model was the cost-sensitive random forest algorithm with a false-negative: false-positive misclassification penalty of 90:1 containing m = 114 predictive features. Model sensitivity and specificity were 0.77 and 0.66, respectively, with an area under the curve of 0.77 in the validation cohort.Conclusions: ML algorithms with resampling and cost-sensitive learning generated clinically valid prediction models for acute desquamation using patient demographic and treatment features. Further external validation and inclusion of genomic markers in ML prediction models are worthwhile, to identify patients at increased risk of toxicity who may benefit from supportive intervention or even a change in treatment plan.",

keywords = "ACUTE SKIN TOXICITY, BIOMARKERS, NORMAL TISSUE, RADIOTHERAPY, RISK-FACTOR",

author = "Mahmoud Aldraimli and Sarah Osman and Diana Grishchuck and Samuel Ingram and Robert Lyon and Anil Mistry and Jorge Oliveira and Robert Samuel and Shelley, {Leila E A} and Daniele Soria and Dwek, {Miriam V} and Aguado-Barrera, {Miguel E} and David Azria and Jenny Chang-Claude and Alison Dunning and Alexandra Giraldo and Sheryl Green and Sara Guti{\'e}rrez-Enr{\'i}quez and Carsten Herskind and {van Hulle}, Hans and Maarten Lambrecht and Laura Lozza and Tiziana Rancati and Victoria Reyes and Rosenstein, {Barry S} and {de Ruysscher}, Dirk and {de Santis}, {Maria C} and Petra Seibold and Elena Sperk and Symonds, {R Paul} and Hilary Stobart and Bego{\~n}a Taboada-Valadares and Talbot, {Christopher J} and Vakaet, {Vincent J L} and Ana Vega and Liv Veldeman and Veldwijk, {Marlon R} and Adam Webb and Caroline Weltens and West, {Catharine M} and Chaussalet, {Thierry J} and Tim Rattay and {REQUITE consortium}",

note = "{\textcopyright} 2022 The Authors.",

year = "2022",

month = jun,

day = "2",

doi = "10.1016/j.adro.2021.100890",

language = "English",

volume = "7",

journal = "Advances in Radiation Oncology",

issn = "2452-1094",

publisher = "Elsevier Inc.",

number = "3",

}

Aldraimli, M, Osman, S, Grishchuck, D, Ingram, S, Lyon, R, Mistry, A, Oliveira, J, Samuel, R, Shelley, LEA, Soria, D, Dwek, MV, Aguado-Barrera, ME, Azria, D, Chang-Claude, J, Dunning, A, Giraldo, A, Green, S, Gutiérrez-Enríquez, S, Herskind, C, van Hulle, H, Lambrecht, M, Lozza, L, Rancati, T, Reyes, V, Rosenstein, BS, de Ruysscher, D, de Santis, MC, Seibold, P, Sperk, E, Symonds, RP, Stobart, H, Taboada-Valadares, B, Talbot, CJ, Vakaet, VJL, Vega, A, Veldeman, L, Veldwijk, MR, Webb, A, Weltens, C, West, CM, Chaussalet, TJ, Rattay, T & REQUITE consortium 2022, 'Development and Optimization of a Machine-Learning Prediction Model for Acute Desquamation After Breast Radiation Therapy in the Multicenter REQUITE Cohort', Advances in Radiation Oncology, vol. 7, no. 3, 100890. https://doi.org/10.1016/j.adro.2021.100890

TY - JOUR

T1 - Development and Optimization of a Machine-Learning Prediction Model for Acute Desquamation After Breast Radiation Therapy in the Multicenter REQUITE Cohort

AU - Aldraimli, Mahmoud

AU - Osman, Sarah

AU - Grishchuck, Diana

AU - Ingram, Samuel

AU - Lyon, Robert

AU - Mistry, Anil

AU - Oliveira, Jorge

AU - Samuel, Robert

AU - Shelley, Leila E A

AU - Soria, Daniele

AU - Dwek, Miriam V

AU - Aguado-Barrera, Miguel E

AU - Azria, David

AU - Chang-Claude, Jenny

AU - Dunning, Alison

AU - Giraldo, Alexandra

AU - Green, Sheryl

AU - Gutiérrez-Enríquez, Sara

AU - Herskind, Carsten

AU - van Hulle, Hans

AU - Lambrecht, Maarten

AU - Lozza, Laura

AU - Rancati, Tiziana

AU - Reyes, Victoria

AU - Rosenstein, Barry S

AU - de Ruysscher, Dirk

AU - de Santis, Maria C

AU - Seibold, Petra

AU - Sperk, Elena

AU - Symonds, R Paul

AU - Stobart, Hilary

AU - Taboada-Valadares, Begoña

AU - Talbot, Christopher J

AU - Vakaet, Vincent J L

AU - Vega, Ana

AU - Veldeman, Liv

AU - Veldwijk, Marlon R

AU - Webb, Adam

AU - Weltens, Caroline

AU - West, Catharine M

AU - Chaussalet, Thierry J

AU - Rattay, Tim

AU - REQUITE consortium

PY - 2022/6/2

Y1 - 2022/6/2

N2 - Purpose: Some patients with breast cancer treated by surgery and radiation therapy experience clinically significant toxicity, which may adversely affect cosmesis and quality of life. There is a paucity of validated clinical prediction models for radiation toxicity. We used machine learning (ML) algorithms to develop and optimise a clinical prediction model for acute breast desquamation after whole breast external beam radiation therapy in the prospective multicenter REQUITE cohort study.Methods and Materials: Using demographic and treatment-related features (m = 122) from patients (n = 2058) at 26 centers, we trained 8 ML algorithms with 10-fold cross-validation in a 50:50 random-split data set with class stratification to predict acute breast desquamation. Based on performance in the validation data set, the logistic model tree, random forest, and naïve Bayes models were taken forward to cost-sensitive learning optimisation.Results: One hundred and ninety-two patients experienced acute desquamation. Resampling and cost-sensitive learning optimisation facilitated an improvement in classification performance. Based on maximising sensitivity (true positives), the "hero" model was the cost-sensitive random forest algorithm with a false-negative: false-positive misclassification penalty of 90:1 containing m = 114 predictive features. Model sensitivity and specificity were 0.77 and 0.66, respectively, with an area under the curve of 0.77 in the validation cohort.Conclusions: ML algorithms with resampling and cost-sensitive learning generated clinically valid prediction models for acute desquamation using patient demographic and treatment features. Further external validation and inclusion of genomic markers in ML prediction models are worthwhile, to identify patients at increased risk of toxicity who may benefit from supportive intervention or even a change in treatment plan.

AB - Purpose: Some patients with breast cancer treated by surgery and radiation therapy experience clinically significant toxicity, which may adversely affect cosmesis and quality of life. There is a paucity of validated clinical prediction models for radiation toxicity. We used machine learning (ML) algorithms to develop and optimise a clinical prediction model for acute breast desquamation after whole breast external beam radiation therapy in the prospective multicenter REQUITE cohort study.Methods and Materials: Using demographic and treatment-related features (m = 122) from patients (n = 2058) at 26 centers, we trained 8 ML algorithms with 10-fold cross-validation in a 50:50 random-split data set with class stratification to predict acute breast desquamation. Based on performance in the validation data set, the logistic model tree, random forest, and naïve Bayes models were taken forward to cost-sensitive learning optimisation.Results: One hundred and ninety-two patients experienced acute desquamation. Resampling and cost-sensitive learning optimisation facilitated an improvement in classification performance. Based on maximising sensitivity (true positives), the "hero" model was the cost-sensitive random forest algorithm with a false-negative: false-positive misclassification penalty of 90:1 containing m = 114 predictive features. Model sensitivity and specificity were 0.77 and 0.66, respectively, with an area under the curve of 0.77 in the validation cohort.Conclusions: ML algorithms with resampling and cost-sensitive learning generated clinically valid prediction models for acute desquamation using patient demographic and treatment features. Further external validation and inclusion of genomic markers in ML prediction models are worthwhile, to identify patients at increased risk of toxicity who may benefit from supportive intervention or even a change in treatment plan.

KW - ACUTE SKIN TOXICITY

KW - BIOMARKERS

KW - NORMAL TISSUE

KW - RADIOTHERAPY

KW - RISK-FACTOR

U2 - 10.1016/j.adro.2021.100890

DO - 10.1016/j.adro.2021.100890

M3 - Article

C2 - 35647396

SN - 2452-1094

VL - 7

JO - Advances in Radiation Oncology

JF - Advances in Radiation Oncology

IS - 3

M1 - 100890

ER -