Comparison of Machine Learning Methods in the Study of Cancer Survivors' Return to Work: An Example of Breast Cancer Survivors with Work-Related Factors in the CONSTANCES Cohort.

PURPOSE: Machine learning (ML) methods showed a higher accuracy in identifying individuals without cancer who were unable to return to work (RTW) compared to the classical methods (e.g. logistic regression models). We therefore aim to discuss the value of these methods in relation to RTW for cancer survivors. METHODS: Breast cancer (BC) survivors who were working at diagnosis within the CONSTANCES cohort were included in the study. RTW was assessed five years after the BC diagnosis (early retirement was considered as non-RTW). Age and occupation at diagnosis, and physical occupational job exposures assessed using the Job Exposure Matrix, JEM-CONSTANCES, were evaluated as predictors of RTW five years after BC diagnosis. The following four ML methods were used: (i) k-nearest neighbors; (ii) random forest; (iii) neural network; and (iv) elastic net. RESULTS: The training sample included 683 BC survivors (RTW: 85.7%), and the test sample 171 (RTW: 85.4%). The elastic net method had the best results despite low sensitivity (accuracy = 76.6%; sensitivity = 31.7%; specificity = 90.8%), and the random forest model was the most accurate (= 79.5%) but also the least sensitive (= 14.3%). CONCLUSION: This study takes a first step towards opening up new possibilities for identifying the occupational determinants of cancer survivors' RTW. Further work, including a larger sample size, and more predictor variables, is now needed.

Development and validation of a work-related risk score for upper-extremity musculoskeletal disorders in a French working population.

OBJECTIVES: The aim was to develop an easy-to-use risk score based on occupational factors and to validate its performance to identify workers either having (diagnostic setting) or developing (prognostic setting) upper-extremity musculoskeletal disorders (UEMSD). METHODS: This study relied on data from the Cosali prospective cohort conducted in a French working population. Diagnostic status for six UEMSD at inclusion and at follow-up was assessed by a standardized clinical examination. Data on occupational factors were collected through a self-administered questionnaire completed before the clinical examination at inclusion. The risk score was derived from a prediction model developed on data of 2,468 workers included in 2002-2003, and the validation sample is composed of 1,051 workers included later in 2004-2005. The prognostic performance of the risk score was assessed in workers without UEMSD at baseline. RESULTS: A total of 13% and 12% of workers had a UEMSD at inclusion in the development and validation sample. The developed risk score includes physical, organizational and psychosocial factors at work. In the validation sample, this score had acceptable performance for identifying workers having or not UEMSD at baseline (AUC: 0.60 [95% CI 0.57 to 0.63]), in particular the negative predictive value was high (89%-90%). The baseline risk score showed similar performance for predicting incident UEMSD at follow-up examination. CONCLUSION: This score can be useful as a first-line risk assessment tool, especially for excluding the low-risk work situations from further intervention by an ergonomist. Further validation studies are needed to determine its performance among various working populations.