Objective   To construct a prediction model for emotional eating behavior during coronavirus disease 2019 (COVID-19) pandemic among doctors in northern region of China for providing evidence to the promotion of healthy dietary patterns in the doctors.

  Methods  An on-site self-administered questionnaire survey was conducted among 2 316 doctors randomly recruited at 39 COVID-19 designated hospitals in 8 provincial-level administrative divions in northern China during May – August 2022. Relevant information of the doctors were collected with a general questionnaire, work-family conflict scale, NEO Five-Factor Inventory and emotional eating scale. Deep neural network (DNN) was used to develop a prediction model for associates of emotional eating during the COVID-19 pandemic in the doctors.

  Results  For 2 094 participants with complete information, the mean overall score of emotional eating during the COVID-19 pandemic was 51.48 ± 17.37 and the dimensional scores were 11.31 ± 4.07 for anger influenced eating, 16.72 ± 7.56 for anxiety influenced eating, 11.02 ± 3.24 for depression influenced eating, and 12.43 ± 4.27 for positive emotion influenced eating. A DNN model with 21-19-14-9-1 network framework was constructed for predicting emotional eating behaviors in the doctors during the COVID-19 pandemic and the parameters of the model were 0.926 for R², 0.039 for mean absolute error, 0.003 for mean squared error, and 0.056 for root mean squared error, respectively. Based on the modeling results, the top five predictors for emotional eating model were alcohol consumption, work-family conflict, unbalanced nutrition, male gender, and irregular dietary pattern.

  Conclusion  Emotional eating behaviors during the COVID-19 pandemic was not rare and mainly influenced by unhealthy eating habits and work-family conflict among doctors in northern China. Machine learning could be used to predict effectively and accurately the risk of emotional eating behavior for the doctors.