Colorectal cancer (CRC) is a prevalent malignant tumor that presents significant challenges to both public health and healthcare systems. The aim of this study was to develop a machine learning model based on five years of clinical follow-up data from CRC patients to accurately identify individuals at risk of poor prognosis. This study included 411 CRC patients who underwent surgery at Yixing Hospital and completed the follow-up process. A modeling dataset containing 73 characteristic variables was established by collecting demographic information, clinical blood test indicators, histopathological results, and additional treatment-related information. Decision tree, random forest, support vector machine, and extreme gradient boosting (XGBoost) models were selected for modeling based on the features identified through recursive feature elimination (RFE). The Cox proportional hazards model was used as the baseline for model comparison. During the model training process, hyperparameters were optimized using a grid search method. The model performance was comprehensively assessed using multiple metrics, including accuracy, F1 score, Brier score, sensitivity, specificity, positive predictive value, negative predictive value, receiver operating characteristic curve, calibration curve, and decision curve analysis curve. For the selected optimal model, the decision-making process was interpreted using the SHapley Additive exPlanations (SHAP) method. The results show that the optimal RFE-XGBoost model achieved an accuracy of 0.83 (95% CI 0.76–0.90), an F1 score of 0.81 (95% CI 0.72–0.88), and an area under the receiver operating characteristic curve of 0.89 (95% CI 0.82–0.94). Furthermore, the model exhibited superior calibration and clinical utility. SHAP analysis revealed that increased perioperative transfusion quantity, higher tumor AJCC stage, elevated carcinoembryonic antigen level, elevated carbohydrate antigen 19–9 (CA19-9) level, advanced age, and elevated carbohydrate antigen 125 (CA125) level were correlated with increased individual mortality risk. The RFE-XGBoost model demonstrated excellent performance in predicting CRC patient prognosis, and the application of the SHAP method bolstered the model’s credibility and utility.

Background and aims: In China, a significant number of undergraduates are experiencing poor sleep quality. This study was designed to investigate the prevalence of poor sleep quality and identify associated factors among undergraduates in Jiangsu Province, China.Methods: A total of 8,457 participants were collected in 2022 using whole-group convenience sampling. The factors studied included basic demographics, family and social support, personal lifestyles, physical and mental health, mobile phone addiction index (MPAI), and the Connor-Davidson resilience scale (CD-RISC). The Pittsburgh Sleep Quality Index (PSQI) was used to assess sleep quality. Four models, including weighted multiple linear regression, binary logistic regression, weighted linear mixed model, and logistic regression with random effects, were applied to identify associated factors for sleep quality.Results: Of the 8,457 participants analyzed, 26.64% (2,253) were classified into the poor sleep quality group with a PSQI score >7. No significant relationship was found between sleep quality and gender, native place, economic level of family, physical exercise, dormitory light, dormitory hygiene, and amativeness matter. Risk factors for sleep quality identified by the four models included lower CD-RISC, higher MPAI, fourth grade or above, smoking, drinking, greater academic pressure, greater employment pressure, roommate sleeping late, noisy dormitory, poorer physical health status, poorer mental health status, and psychological counseling.Conclusions: These findings provide valuable insights for university administrators, enabling them to better understand the risk factors associated with poor sleep quality in undergraduates. By identifying these factors, administrators can provide targeted intervention measures and counseling programs to improve students' sleep quality.

Background: Obesity and metabolic syndrome pose significant health challenges in the United States (US), with connections to disruptions in sex hormone regulation. The increasing prevalence of obesity and metabolic syndrome might be associated with exposure to phthalates (PAEs). Further exploration of the impact of PAEs on obesity is crucial, particularly from a sex hormone perspective.Methods: A total of 7780 adult participants in the National Health and Nutrition Examination Survey (NHANES) from 2013 to 2016 were included in the study. Principal component analysis (PCA) coupled with multinomial logistic regression was employed to elucidate the association between urinary PAEs metabolite concentrations and the likelihood of obesity. Weighted quartiles sum (WQS) regression was utilized to consolidate the impact of mixed PAEs exposure on sex hormone levels (total testosterone (TT), estradiol and sex hormone-binding globulin (SHBG)). We also delved into machine learning models to accurately discern obesity status and identify the key variables contributing most to these models.Results: Principal Component 1 (PC1), characterized by mono(2-ethyl-5-carboxypentyl) phthalate (MECPP), mono(2-ethyl-5-hydroxyhexyl) phthalate (MEHHP), and mono(2-ethyl-5-oxohexyl) phthalate (MEOHP) as major contributors, exhibited a negative association with obesity. Conversely, PC2, with monocarboxyononyl phthalate (MCNP), monocarboxyoctyl phthalate (MCOP), and mono(3-carboxypropyl) phthalate (MCPP) as major contributors, showed a positive association with obesity. Mixed exposure to PAEs was associated with decreased TT levels and increased estradiol and SHBG. During the exploration of the interrelations among obesity, sex hormones, and PAEs, models based on Random Forest (RF) and eXtreme Gradient Boosting (XGBoost) algorithms demonstrated the best classification efficacy. In both models, sex hormones exhibited the highest variable importance, and certain phthalate metabolites made significant contributions to the model’s performance.Conclusions: Individuals with obesity exhibit lower levels of TT and SHBG, accompanied by elevated estradiol levels. Exposure to PAEs disrupts sex hormone levels, contributing to an increased risk of obesity in US adults. In the exploration of the interrelationships among these three factors, the RF and XGBoost algorithm models demonstrated superior performance, with sex hormones displaying higher variable importance.

The prediction of coronavirus disease in 2019 (COVID-19) in broader regions has been widely researched, but for specific areas such as urban areas the predictive models were rarely studied. It may be inaccurate to apply predictive models from a broad region directly to a small area. This paper builds a prediction approach for small size COVID-19 time series in a city. Numbers of COVID-19 daily confirmed cases were collected from November 1, 2022 to November 16, 2023 in Xuzhou city of China. Classical deep learning models including recurrent neural network (RNN), long and short-term memory (LSTM), gated recurrent unit (GRU) and temporal convolutional network (TCN) are initially trained, then RNN, LSTM and GRU are integrated with a new attention mechanism and transfer learning to improve the performance. Ten times ablation experiments are conducted to show the robustness of the performance in prediction. The performances among the models are compared by the mean absolute error, root mean square error and coefficient of determination. LSTM outperforms than others, and TCN has the worst generalization ability. Thus, LSTM is integrated with the new attention mechanism to construct an LSTMATT model, which improves the performance. LSTMATT is trained on the smoothed time series curve through frequency domain convolution augmentation, then transfer learning is adopted to transfer the learned features back to the original time series resulting in a TLLA model that further improves the performance. RNN and GRU are also integrated with the attention mechanism and transfer learning and their performances are also improved, but TLLA still performs best. The TLLA model has the best prediction performance for the time series of COVID-19 daily confirmed cases, and the new attention mechanism and transfer learning contribute to improve the prediction performance in the flatten part and the jagged part, respectively.

Background: Motor symptoms of Parkinson’s disease (PD) patients affect their ability of daily activities. Identifying distinct trajectories of motor symptom progression in PD patients can facilitate long-term management.Methods: A total of 155 PD patients were acquired from the Parkinson’s Disease Progression Marker Initiative (PPMI). Distinct longitudinal trajectory clusters of motor symptom progression in PD patients were identified by unsupervised self-organizing maps (SOMs), and baseline characteristics were compared among different clusters. Linear mixed-effect analysis was utilized to estimate the longitudinal courses of some cardinal motor symptoms among clusters, while survival analysis was used to compare time-to-clinical milestones within 5 years. The support vector machine (SVM) was built to predict patients’ trajectory clusters, and its performance was evaluated through the mean area under the receiver-operating characteristic curve (mAUC), accuracy and macro F₁-score. Shapley values were calculated to interpret individual variability.Results: The optimal clusters by SOMs are 3. Cardinal motor symptoms of the progressive cluster worsened more rapidly, and this cluster is more likely to have impaired balance, loss of independence, sleep disturbance, and cognitive impairment within 5 years. The mAUC, accuracy, and macro F₁-score of multi-class SVM model were 0.8846, 0.7692, and 0.7778, respectively. An interactive web application was developed to predict the individual’s trajectory cluster.Conclusion: Subtyping motor symptom progression into different trajectories can improve patients’ management. Using baseline data to predict which trajectory cluster a patient belongs to may help develop interventions.