Structured assessments of PD lack quantitative measures of functional impairment, remaining highly subjective and variable. Clinicians involved in the long-term management of PD would benefit from a quantitative method to assess patient impairment in the home. To investigate this, nine-axis inertial measurement unit (IMU) sensors were attached to implements used in basic activities of daily living (ADLs), namely brushes and combs. Continuous time-series accelerometer and gyroscope data was collected while people with Parkinson’s Disease (pwPD) performed brushing and combing ADLs. Harnessing temporal and spectral features extracted from time-series data, a linear regression model with L2 regularization achieved a moderately high correlation with patient total score on the Unified Parkinson’s Disease Rating Scale. Additionally, a decision tree machine learning model (XGBoost) trained on signal features accurately performed a classification task, separating participants into healthy control, mild severity PD, and moderate severity PD classes with ~84-88%% accuracy on a held-out test set. Due to the relationship between the patient’s rated severity and features captured by the IMU sensor, this approach has potential use in more objectively characterizing patient functionality longitudinally in the home setting.

Enchante X is a mobile platform that revolutionizes online makeup shopping by combining ultra-realistic AR virtual try-on, AI-driven personalized recommendations, and seamless in-app retail. Users upload a selfie; AI analyzes their skin tone, facial features, and preferences to generate lifelike makeup previews that move naturally with their face. Curated product lists and direct purchase links eliminate guesswork, reduce returns, and boost confidence. With a freemium model, subscription tiers, affiliate commissions, and targeted advertising, Enchante X addresses the $61 billion U.S. beauty-ecommerce market and the $3.7 billion global AI beauty-tech sector. Designed for tech-savvy consumers aged 18–40, GLAMAI delivers truly effortless, engaging, personalized beauty experiences.

Anatomic Endoscopic Enucleation of the Prostate (AEEP) is a highly effective treatment for benign prostatic hyperplasia but remains challenging to master, requiring extensive practice and direct supervision by expert surgeons. This dependence limits opportunities for independent learning and slows trainee progression. To address this, we developed a hydrogel-based simulation model that replicates male pelvic anatomy, including the prostate and bladder. The model incorporates real-time force feedback using force sensors connected to a Raspberry Pi and Arduino. Trainees receive live visual cues, green (safe), yellow (near-threshold), and red (excessive), to guide safe and effective instrument use. An IMU sensor attached to the endoscope handle enables a dynamic minimap that tracks the tool’s position within the model. After the procedure, a feedback dashboard provides performance analytics for self-guided improvement. Our system promotes independent learning and enables efficient use of expert surgeon time, ultimately enhancing technical skill acquisition and clinical decision-making through simulation-based education.

Sepsis is a life-threatening condition characterized by a dysregulated host response to infection. Early prediction is crucial for improving outcomes but remains a challenge using traditional models. Leveraging data from 11,325 adult ICU patients in the MIMIC-III dataset, we developed a transformer-based model to predict sepsis onset and outcomes using clinical features and high-frequency physiological waveforms (e.g., ECG, PPG, ABP). Clinical and waveform data were tokenized into learnable embeddings, enabling the model to capture temporal dynamics through cross-attention mechanisms. Performance evaluation showed strong predictive ability at shorter lead times, with reduced accuracy at longer horizons, highlighting the need for richer input features. To address this, we built a waveform tokenization pipeline to distill informative segments from raw signals. Token importance analysis revealed distinct physiological and clinical markers associated with sepsis. Future work will focus on fusing clinical and waveform embeddings to improve accuracy and interpretability for early, actionable sepsis prediction in critical care.