EnerSys (United States)

Personalized healthcare, driven by lifestyle-informed prediction, has the potential to revolutionize preventive medicine by enabling early detection of health risks. This study proposes an integrated framework that combines machine learning (ML), deep learning (DL), and explainable AI (XAI) for lifestyle-driven health classification. The system processes lifestyle and health attributes including exercise frequency, body mass index (BMI), dietary habits, and sleep duration through a modular pipeline of preprocessing, predictive modeling, explainability, and evaluation. Experimental results on the curated dataset demonstrate that deep learning models significantly outperform traditional ML baselines. The LSTM achieved the best performance with an accuracy of 92.4%, F1-score of 0.91, and AUC of 0.95, while the DNN followed closely with 90.8% accuracy. In contrast, traditional models such as Logistic Regression and Random Forest achieved 82.3% and 85.6% accuracy, respectively. Beyond predictive performance, explainability analysis via SHAP identified exercise frequency, BMI, and sleep duration as the most influential global predictors, while LIME provided instance-level explanations, revealing how high physical activity mitigates risks associated with elevated BMI. These results highlight the dual advantage of robust predictive modeling and transparent interpretation. The proposed framework contributes to advancing trustworthy, personalized, and preventive healthcare, bridging the gap between accuracy and interpretability in clinical decision support.

Motive power battery charger needs to survive in a harsh industrial environment. A typical charger power converter module today utilizes direct forced air cooling, which introduces corrosive damages to the power converter components. This paper presents an innovative design for SiCbased LLC resonant converter with all natural convection cooling operating in wide voltage range application. In this design, there is no corrosive cooling air touching the power converter components. Therefore, the charger modules are much more durable. Both the phase shift full bridge control and the half and full bridge mode transition control are proposed to regulate the wide voltage range. The voltage gain of LLC resonant converter of both half bridge and full bridge is analyzed. The performance of the proposed topologies is verified through mathematical analysis, simulation and experimental results.

Energy storage systems have supported the Telecom, Utility, and Data Center markets for nearly 100 years. This paper will review the traditional storage systems and technologies by using matrices which characterize the product solutions benefits and weaknesses. Other critical characteristics such as energy density, floor loading, will be compared and contrasted. Examples of new system architectures will be reviewed and assessed for benefits and shortcomings including a TCO model.

Hurricane Katrina and the pursuant ruling by the FCC has caused many Telecom operators to review and find solutions to provide extended runtime in critical areas of their network. In the past, lead acid batteries have dominated the backup power market for these sites and the power grid quality was considered robust. Katrina forced operators to look past this and it revealed that lead acid batteries typically only provide up to eight hours of backup for a site. For those sites where operators require extended runtime, a diesel generator was typically the only option. This paper will provide an overview of not only the diesel solution but also for a new solution that utilizes a proton exchange membrane fuel cell (PEMFC) and premium lead acid batteries to provide extended runtime at the site using a 5 kW base station plant, we will examine the architecture and design solutions that will give operators well over 48 hours of backup and in some cases over 160 hours. The PEMFC solution can utilize advanced lead acid front terminal batteries to ease maintenance as well as a robust and cost effective PEMFC engine. Critical areas for the mainstream success of fuel cells include total solution costs, reliability, fuel supply, maintenance and hydrogen storage. The comparative diesel will be the embedded 30 kW solution that is normally found in sites throughout these applications. Issues associated with generator operation include noise, emissions, permitting, fuel supply and maintenance costs will all be examined. The goal will be to show a total cost of ownership for both solutions and educate operators about new approaches for their extended runtime requirements.