Internet-based Cognitive Behavioral Therapy (ICBT) provides online platforms to patients facing psychiatric disorders with strategies for coping with mental health challenges. The application of machine learning for treatment outcome prediction that can assess whether the ongoing therapy will succeed or fail by the end of treatment is crucial for optimizing the operations of such ICBT providers. This paper presents Deep Neural Decision Forests (DNDFs) to process tabular ICBT data for clinical outcome prediction. We apply a neural network to learn a meaningful linear transformation from multiview tabular data's mixed feature types (e.g., nominal, ordinal). Then, we pass it through a forest of differentiable trees supporting a stochastic routing mechanism. The average class-specific probability scores learned through such a decision forest depict the final clinical outcome. We conduct experiments using de-identified data from an ICBT company to evaluate the learned patient representations for treatment outcome prediction. DNDFs significantly excel all baseline approaches for studied multiview tabular data.

Crime Scene Investigation (CSI) is a forensic crime-based series where perpetrators often try to hide their motives to cover up murders. In contrast, investigators trace pieces of evidence to spot culprits. Recognizing the original character played by a particular speaker (i.e., perpetrator, investigator, and suspects), corresponding to any CSI-based dialogue, using textual conversations is challenging. Existing approaches do not use deep multiview learning for processing multiview commonsense-based Knowledge Graph (KG). Our proposed approach, RiMCR, first applies Siamese BERT-Networks (SBERT) to learn sentence structure. We process sixteen multiview relations of commonsense-based knowledge graph ATOMIC₂₀²⁰ through COMET(BART). A dual-view deep network architecture based on independent stacked LSTMs with a self-attention mechanism infuses sequential patterns into sentence and common-sense-based features. Lastly, we concatenate four types of encoded features before passing through the decoder to solve binary and multiclass classification problems. An extensive comparison with sequence models and Large Language Models (LLMs) validates the judiciousness of RiMCR.

A careful analysis of the Large Language Model (LLM) results, generated through anonymized representations of the original dataset, is crucial to precisely evaluate the data-sharing procedure's limitations and facilitate valuable collaborations among Internet-based cognitive behavioral therapy (ICBT) companies and third parties. This paper presents an experimental study of fine-Tuning 27 LMs for a multiclass classification task to identify depression severity using 40,191 tweets labeled by human annotators. We fine-Tune 14 Bidirectional Encoder Representations from Transformers (BERT), 6 Robustly Optimized BERT Pretraining Approaches (RoBerta), 3 Generative Pretraining (GPT), and 4 Text-To-Text Transfer Transformer (T5) based LMs to classify confidential and anonymized tweets. We report that T5, through conditional generation, outperforms widely adopted BERT, RoBerta, and GPT types for classifying confidential and anonymized tweets. Anonymizing personal information safeguards user privacy and often increases LM performance. Case sensitivity can potentially improve or harm the performance of domain-specific LMs for original and anonymized text.

Crime forecasting techniques can play a leading role in hindering crime occurrences, especially in areas under possible threat. In this paper, we propose Locality Aware Temporal Factorization Machines (LTFMs) for crime prediction. Its locality representation module deploys a spatial encoder to estimate the regional dependencies using Graph Convolutional Networks (GCNs). Then, the Point of Interest (POI) encoder computes the weighted attentive aggregation of location, crime, and POI latent representations. The dynamic crime representation module utilizes the transformer-based positional encodings to capture the dependencies among space, time, and crime categories. The encodings learnt from locality representation and crime category encoders, are projected into a factorization machine-based architecture via a shared feed-forward network. An extensive comparison with state-of-art techniques, using Chicago and New York's criminal records, shows the significance of LTFMs.