In a recent development, Google’s Personal Health Large Language Model (PH-LLM), a specialized version of its Gemini LLM, has demonstrated superior performance in providing sleep and fitness advice compared to human experts. The PH-LLM, fine-tuned to analyze and interpret time-series data from wearable devices, was tested on its ability to answer health-related questions and make accurate predictions. Researchers found that the model consistently outperformed experienced professionals in the health and fitness domain, showcasing its potential to enhance personal health management through advanced AI technology.

Google Gemini proves a better health coach than humans – RamaOnHealthcare VentureBeat

### Study: AI-powered Alerts Improve Patient Outcomes (June 14, 2024)
A study conducted at Mount Sinai Hospital, published in Critical Care Medicine, revealed that AI-generated alerts significantly enhance patient care and outcomes. Patients whose care teams received real-time AI alerts about health deterioration were 43% more likely to get timely escalated care and had a lower mortality rate within 30 days. The research led by Dr. Matthew Levin demonstrates that AI-driven alerts outperform traditional methods in predicting clinical decline, facilitating earlier interventions and saving lives. Although the study faced interruptions due to the COVID-19 pandemic, the algorithm has been widely implemented across Mount Sinai's stepdown units, aiding clinical decision-making and improving patient safety through continuous data integration and reinforcement learning.

Both the House and Senate Armed Services Committees have passed measures in their respective versions of the National Defense Authorization Act (NDAA) calling for independent assessments on the potential establishment of a U.S. Cyber Force. This move aims to address the increasing cyber threats and current fragmented approaches within military services, which have led to readiness issues. The Senate had a similar provision last year, but it was removed during reconciliation with the House. The current effort includes an amendment proposed by Rep. Morgan Luttrell, directing the National Academies of Sciences, Engineering, and Medicine to conduct the study. Proponents argue that each service's isolated cyber capacity building is insufficient, and the need for a unified cyber military service is critical. The Department of Defense (DOD) is currently bundling various required studies under the "Cybercom 2.0" effort to address future cyber force structuring, with reports due in June 2025. Despite the push for new assessments, internal resistance within the DOD suggests a challenging path ahead for the proposed independent cyber force.

Assessment for independent cyber force passes House, Senate defense committee defensescoop

The article "7 Paradoxes of Artificial Intelligence" by Joe McKendrick explores seven conflicting aspects of AI technology that complicate its implementation for business and IT leaders. These paradoxes include: AI reducing labor needs while increasing skill requirements, being complex yet simplifying application development, incurring high costs but aiding cost management, automating work while demanding creativity, benefiting advanced companies more than struggling ones, requiring large datasets yet alleviating data management, and possessing significant intelligence but failing at simple tasks. These contradictions highlight the nuanced and perplexing nature of AI adoption in the business world.

7 Paradoxes Of Artificial Intelligence Forbes