Sentiment analysis is crucial in healthcare to understand patients’ emotions, automatically identifying the feelings of patients suffering from serious illnesses (cancer, AIDS, or Ebola) with an artificial intelligence model that constitutes a major challenge to help health professionals. This study presents a comparative study on different machine learning (logistic regression, naive Bayes, and LightGBM) and deep learning models: long short-term memory (LSTM) and bidirectional encoder representations from transformers (BERT) for classify health feelings thanks to textual data related to patients with serious illnesses. Considering the class imbalance of the dataset, various resampling techniques are investigated. The approach is complemented by an explainable model, LIME, to understand the shortcomings of the classification results. The results highlight the superior performance of the BERT and LSTM models with an F1-score of 89%.
Multimodal behavioral sensing may support depression screening, but evaluation on small clinical-interview datasets is particularly vulnerable to data leakage and model-selection bias. We present a leakage-audited trimodal framework evaluated on DAIC-WOZ (, PHQ-8 ), combining SBERT text embeddings, OpenFace facial-behavior descriptors, and COVAREP prosodic features. Participant-level partitioning is performed before augmentation, while decision thresholds and neural-model checkpoints are selected exclusively from internal validation data. A controlled five-seed experiment showed that a deliberately leaky full-pool MixUp construction, in which a retained development sample could include a held-out participant as its second parent, was associated with a 27–33 percentage-point increase in Macro-F1 across four fusion configurations. Under the participant-level leakage-free 5-fold protocol, trimodal late fusion achieved a Macro-F1 of , compared with for Text+Imaging late fusion. Paired participant-level correctness outcomes also favored trimodal fusion (McNemar , ), consistent with improved paired classification when the audio modality was included under the leakage-free protocol. On 86 participant-disjoint E-DAIC sessions, using a consistent PHQ-8-based outcome definition (PHQ-8 ), trimodal late fusion achieved Macro-F1 = 0.621 and AUC = 0.676; this experiment is interpreted as within-family generalization rather than independent cross-corpus validation. Overall, the results show that leakage control can substantially alter both absolute performance and comparative conclusions, and that multimodal gains should be established through participant-level evaluation, modality-specific analysis, and reproducible model-selection procedures.