PurposeComputer algorithms and Machine Learning (ML) will be integrated into clinical decision support within occupational health care. This will change the interaction between health care professionals and their clients, with unknown consequences. The aim of this study was to explore ethical considerations and potential consequences of using ML based decision support tools (DSTs) in the context of occupational health.MethodsWe conducted an ethical deliberation. This was supported by a narrative literature review of publications about ML and DSTs in occupational health and by an assessment of the potential impact of ML-DSTs according to frameworks from medical ethics and philosophy of technology. We introduce a hypothetical clinical scenario from a workers’ health assessment to reflect on biomedical ethical principles: respect for autonomy, beneficence, non-maleficence and justice.ResultsRespect for autonomy is affected by uncertainty about what future consequences the worker is consenting to as a result of the fluctuating nature of ML-DSTs and validity evidence used to inform the worker. A beneficent advisory process is influenced because the three elements of evidence based practice are affected through use of a ML-DST. The principle of non-maleficence is challenged by the balance between group-level benefits and individual harm, the vulnerability of the worker in the occupational context, and the possibility of function creep. Justice might be empowered when the ML-DST is valid, but profiling and discrimination are potential risks.ConclusionsImplications of ethical considerations have been described for the socially responsible design of ML-DSTs. Three recommendations were provided to minimize undesirable adverse effects of the development and implementation of ML-DSTs.

Low heart rate variability (HRV) is related to health problems that are known reasons for sick-leave or early retirement. A 1-minute-protocol could allow large scale HRV measurement for screening of health problems and, potentially, sustained employability. Our objectives were to explore the association of HRV with measures of health. Cross-sectional design with 877 Dutch employees assessed during a Workers’ Health Assessment. Personal and job characteristics, workability, psychological and mental problems, and lifestyle were measured with questionnaires. Biometry was measured (BMI, waist circumference, blood pressure, glucose, cholesterol). HRV was assessed with a 1-minute paced deep-breathing protocol and expressed as mean heart rate range (MHRR). A low MHRR indicates a higher health risk. Groups were classified age adjusted for HRV and compared. Spearman correlations between raw MHRR and the other measures were calculated. Significant univariable correlations (p < 0.05) were entered in a linear regression model to explore the multivariable association with MHRR. Age, years of employment, BMI and waist circumference differed significantly between HRV groups. Significant correlations were found between MHRR and age, workability, BMI, waist circumference, cholesterol, systolic and diastolic blood-pressure and reported physical activity and alcohol consumption. In the multivariable analyses 21.1% of variance was explained: a low HRV correlates with aging, higher BMI and higher levels of reported physically activity. HRV was significantly associated with age, measures of obesity (BMI, waist circumference), and with reported physical activity, which provides a first glance of the utility of a 1-minute paced deep-breathing HRV protocol as part of a comprehensive preventive Workers’ Health Assessment.

Objective: The aging workforce influences employability and health of the working population, with new challenges emerging. The focus has shifted from return to work only, to enhancing ability to stay at work. It is unclear whether factors that influence return to work (RTW) also apply to preserving health and helping workers stay at work (SAW). Study objectives were to identify factors contributing to SAW among industrial workers and map identified factors to the Arena of Work Disability Prevention model (WDP-Arena, a commonly used RTW model) to identify agreements and differences. Methods: Scoping review; eight databases were searched between January 2005- January 2020. Manuscripts with SAW as outcome were included; manuscripts with (early) retirement as outcome were excluded. Factors contributing to SAW were mapped against the components of the WDP-Arena. Results: Thirteen manuscripts were included. Most results aligned with the WDP-Arena. These were most often related to the Workplace and Personal system. Compared to RTW, in industrial workers fewer factors related to the Legislative and Insurance system or Health Care system were relevant for SAW. Societal/cultural/political context was not studied. Multidimensional factors (workability, vitality at work, balanced workstyle, general health, dietary habits) were related to SAW but did not align with components in the WDP-Arena. Conclusion: Most factors that determine SAW in industrial workers could be mapped onto the WDP- Arena model. However, new influencing factors were found that could not be mapped because they are multidimensional. The life-course perspective in SAW is more evident than in RTW. Many elements of the Legislative and Insurance system and the Health Care system have not been studied.

BACKGROUND: Sustainable employability is increasingly important with current socio-economic challenges. Screening for resilience could contribute to early detection of either a risk, or a protector for sustainable employability, the latter being operationalized as workability and vitality. OBJECTIVE: To study the predictive value of Heart Rate Variability (HRV) measurements and the Brief Resilience Scale (BRS) for worker self-reported workability and vitality after 2–4 years. METHODS: Prospective observational cohort study with mean follow-up period of 38 months. 1,624 workers (18–65 years old) in moderate and large companies participated. Resilience was measured by HRV (one-minute paced deep breathing protocol) and the BRS at baseline. Workability Index (WAI), and the Vitality dimension of the Utrecht Work Engagement Scale-9 (UWES-9-vitality) were the outcome measures. Backward stepwise multiple regression analysis (p < 0.05) was performed to evaluate the predictive value of resilience for workability and vitality, adjusted for body mass index, age and gender. RESULTS: N = 428 workers met inclusion criteria after follow-up. The contribution of resilience, measured with the BRS, was modest but statistically significant for the prediction of vitality (R2 = 7.3%) and workability (R2 = 9.2%). HRV did not contribute to prediction of workability or vitality. Age was the only significant covariate in the WAI model. CONCLUSION: Self-reported resilience modestly predicted workability and vitality after 2–4 years. Self-reported resilience may provide early insight into the ability of workers to stay at work, although caution must be applied because explained variance was modest. HRV was not predictive.

Background: Occupational Health Check-ups (OHCs) have evolved from identifying work-related health risks to supporting sustainable employability through individualized lifestyle advice, but translating multidimensional demographic, occupational, and health characteristics into effective, personalized recommendations remains complex and lacks validated decision-support methods. Objective: Aims of this study are to: 1) develop a Machine Learning Decision Support Tool (ML-DST) based on workers’ demographic, occupational, and health characteristics to support personalized advice about lifestyle changes to improve sustainable employability, and 2) evaluate performance of the ML-DST. Methods: Real-world OHC data from a population-based cohort (2010–2020) were used to develop ML-DST models providing lifestyle advice (physical activity, smoking, alcohol, nutrition, relaxation, or no change) to improve workability and vitality. Paired baseline and 1–4-year follow-up data were used to train and evaluate 12 models. Per model, three rule-based algorithms were applied. Performance was considered useful when both True Positive Rate (TPR) and macroF1 score were ≥0.7. The best performing algorithm is proposed per model for inclusion in the final ML-DST. Results: Of 20,440 individual OHCs, 3412 paired sets were included. Best TPRs ranged from 0.51 to 0.71. One model achieved TPR and macroF1 scores ≥0.7, supporting nutritional behavior change to improve vitality; however, its profiles were not conceptually meaningful. The other 11 models did not achieve useful quantitative performance. Conclusion: Training algorithms for an ML-DST to support advice about lifestyle changes to improve long-term sustainable employability using data from regular Occupational Health Check-ups did not lead to useful model performance for improving workability and vitality.