• April 2026
• Article

The EMOTE-TNK safety and tolerability study calculated financial savings from early mobilization with tenecteplase (TNK) of acute ischemic stroke patients. The study used the Time-Driven Activity-Based Costing methodology to compare costs of 180 patients, treated between February 2021 and March 2024, who had early mobilization, defined as occurring between 13 and 24 hours after thrombolysis, with those of patients who had conventional mobilization, defined as waiting 24 hours after thrombolysis. By starting rehabilitation service evaluations earlier, the average duration of hospitalization was reduced by 0.5 days and treatment costs dropped by $1,250 per patient, a 25% saving.

• 2026
• Article

Large language models (LLMs) have emerged as potential tools for early disease characterization and pandemic preparedness due to their ability to interpret complex textual data. This study evaluated the sensitivity of three LLMs: GPT-5, Claude Sonnet 4, and Gemini 2.5 Pro on early case reports of COVID-19, Mpox, and Zika. Each case report was modified to remove explicit diagnostic terms, and models were prompted to identify whether the presentation represented a disease of pandemic potential. Claude Sonnet 4 achieved the highest sensitivity overall across all three diseases. GPT-5 demonstrated inconsistent results, performing poorly on Mpox. Findings highlight significant variability in diagnostic reliability across LLMs, emphasizing the need for multimodal integration, dataset refinement, and ethical oversight. Limitations include the small sample size, retrospective English-language case reports, text-only inputs, and evaluation of known diseases.

• 2026
• Working Paper

Importance: During past public health crises, from mass casualty events to the COVID-19 pandemic, countless patients suffered needless morbidity and mortality because the availability of nearby, crucially needed resources was not visible to over-capacity providers and other stakeholders. Residents of rural areas, lower income communities, Black patients, and those with chronic conditions suffered disproportionately. Data about reduced availability of needed resources were transmitted irregularly to the federal government. Governments and many sites of care lacked the analytic tools to create coordinated resource allocation at the local level. Future crisis situations and the increasing shortage of hospital beds require that sites of care and public health entities abate unnecessary morbidity and mortality by improving the alignment of capacity with projected demand. Observations: Systems exist for real time data transmission, artificial intelligence-driven predictive modeling, and real-time resource allocation. They can anticipate demand and rapidly direct patient flow and redistribute critical care resources to prevent overwhelming individual sites of care. They harmonize disparate data in a coherent, actionable, local resource management framework with criteria such as local capacity and patient acuity. A counterfactual simulation framework, leveraging artificial intelligence-driven optimization to estimate the impact of inter-hospital transfers under a transparent data environment, conservatively estimated a mortality decrease of 3-5% that could now avoid a detrimental load imbalance for the more than 400,000 patient arrivals at the hospitals studied. Conclusions and Relevance: Federal government requirements for real-time disclosure of resource data and artificial intelligence modeling for emergency public health medical care can reduce the morbidity and mortality that occurred when healthcare entities faced sudden, substantial demands for critically needed resources. These measures can also enable internal hospital quality and efficiency innovations to control costs, improve access, and reduce inequity. As in prior requirements for disclosure, existing federally mandated incentives or penalties and mechanisms for assuring data collection and implementation of incentives or penalties can apply to these requirements. The data, their transmission, the incentives, and the artificial intelligence models should be routinely assessed and adjusted to ensure their effectiveness, equity, and compliance with up-to-date standards.

• February 5, 2026
• Article

The corporatization of the U.S. health insurance industry may contribute to poor health care performance for the commercially insured population, but some supply-side and demand-side reforms could help.

• February 2026 (Revised March 2026)
• Case

In the fall of 2022, Blackstone's private equity team was considering a potential corporate carve-out deal for its flagship buyout fund series. The deal centered around Copeland, a business unit within Emerson Electric that produced a leading share of compressors used in residential and commercial HVACs. Rather than run a traditional process, Emerson entered formal exclusivity with Blackstone due to the complex nature of the transaction and significant upfront cash equity commitment. Copeland benefitted from changes in the regulatory environment and increased demand brought on by rising temperatures and the COVID-era surge in work from home. However, the deal came with risks: the macroeconomic environment was uncertain and the growth profile of Copeland was fairly modest, as the surge in HVAC replacements in the wake of the pandemic was unlikely to persist. The Blackstone team was left to evaluate whether operational improvements and the structure of the deal were sufficient to mitigate potential risks. This case provides an opportunity to dig into a complex, bilateral, carve-out in the context of a unique macroeconomic backdrop.

• February 2026 (Revised March 2026)
• Case

This case describes how Mass General Brigham (MGB), the largest health system in Massachusetts, piloted an AI-powered ambient documentation program among a subset of clinicians. Renowned for its strong emphasis on research, innovation, and patient health outcomes, clinicians at MGB were also experiencing increasing rates of burnout and turnover, largely driven by the administrative burden of clinical documentation in the electronic health records (EHR). To address this issue, they introduced a pilot program to test ambient documentation tools which used natural language processing to transcribe and summarize physician-patient encounters, reducing time spent on documentation and allowing physicians to focus more on patient care. Early results showed promising reductions in burnout, intention to leave, and increased physician efficiency, but concerns about the technology remained. The technology could generate inaccurate notes with errors or hallucinations which made physicians resistant to learn and adopt it into their workflows. Leadership contemplated how to scale this technology responsibly without exacerbating existing challenges.