Clinical trial matching is the task of identifying trials for which patients may be eligible. Typically, this task is labor-intensive and requires detailed verification of patient electronic health records (EHRs) against the stringent inclusion and exclusion criteria of clinical trials. This process also results in many patients missing out on potential therapeutic options. Recent advancements in Large Language Models (LLMs) have made automating patient-trial matching possible, as shown in multiple concurrent research studies. However, the current approaches are confined to constrained, often synthetic, datasets that do not adequately mirror the complexities encountered in real-world medical data. In this study, we present an end-to-end large-scale empirical evaluation of a clinical trial matching system and validate it using real-world EHRs. We perform comprehensive experiments with proprietary LLMs and our custom fine-tuned model called OncoLLM and show that OncoLLM outperforms GPT-3.5 and matches the performance of qualified medical doctors for clinical trial matching.

Released in November of 2022 as a public and free research preview, ChatGPT rapidly brought OpenAI's work with autoregressive large language modeling based on Generative Pretrained Transformers (GPT) into the mainstream. 1 Users from around the world were able to easily interact with the ChatGPT model through a simple web interface (chat.openai.com). This model has now been used to generate scholarly content, take board exams, and write discharge summaries. 2,3 Less than 2 months from public launch, Chat-GPT had more than 100 million active users and OpenAI. com became a top 50 most-visited websites in the world. 4 The best explanation about why there was such rapid adoption and popularity might be directly from the source: Author Prompt: "Why did ChatGPT become so popular?" ChatGPT Response: "As an AI language model, ChatGPT became popular because of its impressive ability to understand and generate human-like language. It was trained on a

PURPOSE: The COVID-19 pandemic has resulted in significant changes in health care delivery, including the rapid adoption of telemedicine across multiple specialties and practice environments. This includes postoperative visits (POV), despite limited data on outcomes following these telemedicine POV. We sought to determine whether these types of visits successfully identify and address postoperative complications when compared with in-person POV. METHODS: This was a retrospective cohort study of patients undergoing elective inpatient cancer-related surgery from March 2020 through December 2020. The exposure variable was type of POV (telemedicine v in-person). The primary outcome was unplanned hospital readmission within 90 days, and secondary outcomes included 30-day readmission, length of stay of first readmission, and mortality. RESULTS: Five-hundred thirty-five patients underwent elective inpatient operations and met our inclusion criteria. Of these, 98 (18.5%) had an initial telemedicine POV. There was no difference in 90-day readmission on the basis of POV type (16.3% telemedicine v 16.5% in-person, P = .99). Reasons for readmission did not differ between patients who underwent a telemedicine POV compared with in-person POV (all P > .05). After adjustment for patients' demographic and clinical factors, telemedicine POV was not associated with 90-day readmission (odds ratio, 0.89; 95% CI, 0.43 to 1.70; P = .77). CONCLUSION: Telemedicine POV use adopted during the COVID-19 pandemic did not increase risk of readmission when compared with in-person visits following inpatient oncologic surgery. These data can help inform policy on the continued use and application of telemedicine after the pandemic.

Early postoperative mortality risk prediction is crucial for clinical management of gastric cancer. This study aims to predict 90-day mortality in gastric cancer patients undergoing gastrectomy using automated machine learning (AutoML), optimize models for preoperative prediction, and identify factors influential in prediction. National Cancer Database was used to identify stage I–III gastric cancer patients undergoing gastrectomy between 2004 and 2016. 26 features were used to train predictive models using H2O.ai AutoML. Performance on validation cohort was measured. In 39,108 patients, 90-day mortality rate was 8.8%. The highest performing model was an ensemble (AUC = 0.77); older age, nodal ratio, and length of inpatient stay (LOS) following surgery were most influential for prediction. Removing the latter two parameters decreased model performance (AUC 0.71). For optimizing models for preoperative use, models were developed to first predict node ratio or LOS, and these predicted values were inputted for 90-day mortality prediction (AUC of 0.73–0.74). AutoML performed well in predicting 90-day mortality in a larger cohort of gastric cancer patients that underwent gastrectomy. These models can be implemented preoperatively to inform prognostication and patient selection for surgery. Our study supports broader evaluation and application of AutoML to guide surgical oncologic care.