A systematic evaluation of the performance of GPT-4 and PaLM2 to diagnose comorbidities in MIMIC-IV patients.

Background: Given the strikingly high diagnostic error rate in hospitals, and the recent development of Large Language Models (LLMs), we set out to measure the diagnostic sensitivity of two popular LLMs: GPT-4 and PaLM2. Small-scale studies to evaluate the diagnostic ability of LLMs have shown promising results, with GPT-4 demonstrating high accuracy in diagnosing test cases. However, larger evaluations on real electronic patient data are needed to provide more reliable estimates. Methods: To fill this gap in the literature, we used a deidentified Electronic Health Record (EHR) data set of about 300,000 patients admitted to the Beth Israel Deaconess Medical Center in Boston. This data set contained blood, imaging, microbiology and vital sign information as well as the patients' medical diagnostic codes. Based on the available EHR data, doctors curated a set of diagnoses for each patient, which we will refer to as ground truth diagnoses. We then designed carefully-written prompts to get patient diagnostic predictions from the LLMs and compared this to the ground truth diagnoses in a random sample of 1000 patients. Results: Based on the proportion of correctly predicted ground truth diagnoses, we estimated the diagnostic hit rate of GPT-4 to be 93.9%. PaLM2 achieved 84.7% on the same data set. On these 1000 randomly selected EHRs, GPT-4 correctly identified 1116 unique diagnoses. Conclusion: The results suggest that artificial intelligence (AI) has the potential when working alongside clinicians to reduce cognitive errors which lead to hundreds of thousands of misdiagnoses every year. However, human oversight of AI remains essential: LLMs cannot replace clinicians, especially when it comes to human understanding and empathy. Furthermore, a significant number of challenges in incorporating AI into health care exist, including ethical, liability and regulatory barriers.

Mechanism design for a fair and equitable approach to global vaccine distribution: The case of COVID-19.

Vaccines are one of the most effective tools humanity has in the fight against pandemics. One of the major challenges of vaccine distribution is achieving fair and equitable allocation across the countries of the world, regardless of their economic wealth. The self-interested behaviour of high-income countries and the underutilisation of vaccines allocated to underprepared countries are some of the failures reported during COVID-19 vaccine roll-out. These shortcomings have motivated the need for a central market mechanism that takes into account the countries' vulnerability to COVID-19 and their readiness to distribute and administer their allocated vaccines. In this paper, we leverage game theory to study the problem of equitable global vaccine distribution and propose a fair market mechanism that aligns self-interested behaviour with optimal global objectives. First, we model the interaction between a central vaccine provider (e.g. COVAX) and a country reporting its demand as a two-player game, and discuss the Nash and mixed Nash equilibria of that game. Then, we propose a repeated auction mechanism with an artificial payment system for allocating vaccines among participating countries, where each auction round is based on a Vickrey-Clarke-Groves (VCG) mechanism. The proposed allocation mechanism aims at minimising deaths and incentivises the self-interested countries to report their demand truthfully. Compared with real-world COVAX allocation decisions, our results show that the proposed auction mechanism achieves more efficient outcomes that maximise the number of averted deaths. Pragmatic considerations are investigated and policy recommendations are discussed.

Game theory to enhance stock management of Personal Protective Equipment (PPE) during the COVID-19 outbreak.

Since the outbreak of the COVID-19 pandemic, many healthcare facilities have suffered from shortages in medical resources, particularly in Personal Protective Equipment (PPE). In this paper, we propose a game-theoretic approach to schedule PPE orders among healthcare facilities. In this PPE game, each independent healthcare facility optimises its own storage utilisation in order to keep its PPE cost at a minimum. Such a model can reduce peak demand considerably when applied to a variable PPE consumption profile. Experiments conducted for NHS England regions using actual data confirm that the challenge of securing PPE supply during disasters such as COVID-19 can be eased if proper stock management procedures are adopted. These procedures can include early stockpiling, increasing storage capacities and implementing measures that can prolong the time period between successive infection waves, such as social distancing measures. Simulation results suggest that the provision of PPE dedicated storage space can be a viable solution to avoid straining PPE supply chains in case a second wave of COVID-19 infections occurs.

Game On? Smoking Cessation Through the Gamification of mHealth: A Longitudinal Qualitative Study.

BACKGROUND: Finding ways to increase and sustain engagement with mHealth interventions has become a challenge during application development. While gamification shows promise and has proven effective in many fields, critical questions remain concerning how to use gamification to modify health behavior. OBJECTIVE: The objective of this study is to investigate how the gamification of mHealth interventions leads to a change in health behavior, specifically with respect to smoking cessation. METHODS: We conducted a qualitative longitudinal study using a sample of 16 smokers divided into 2 cohorts (one used a gamified intervention and the other used a nongamified intervention). Each participant underwent 4 semistructured interviews over a period of 5 weeks. Semistructured interviews were also conducted with 4 experts in gamification, mHealth, and smoking cessation. Interviews were transcribed verbatim and thematic analysis undertaken. RESULTS: Results indicated perceived behavioral control and intrinsic motivation acted as positive drivers to game engagement and consequently positive health behavior. Importantly, external social influences exerted a negative effect. We identified 3 critical factors, whose presence was necessary for game engagement: purpose (explicit purpose known by the user), user alignment (congruency of game and user objectives), and functional utility (a well-designed game). We summarize these findings in a framework to guide the future development of gamified mHealth interventions. CONCLUSIONS: Gamification holds the potential for a low-cost, highly effective mHealth solution that may replace or supplement the behavioral support component found in current smoking cessation programs. The framework reported here has been built on evidence specific to smoking cessation, however it can be adapted to health interventions in other disease categories. Future research is required to evaluate the generalizability and effectiveness of the framework, directly against current behavioral support therapy interventions in smoking cessation and beyond.