Citations

Abstract Endometriosis is a chronic painful condition affecting 10% of reproductive-aged women with a high prevalence of overlapping gastrointestinal symptoms. There is a significant diagnostic delay for endometriosis and current screening practices for endometriosis in gastroenterology clinic populations have not been evaluated. 112 gastroenterology providers (57.5% females, 64.3% physicians) completed an anonymized exploratory survey (40.1% response rate). Most respondents (59.4%) currently do not screen for endometriosis, but the majority (72.1%) reported they would be “extremely likely” or “somewhat likely” to screen for endometriosis after being presented with data on the high prevalence of chronic GI symptoms in patients with endometriosis.

Abstract An accurate assessment of p53’s functional statuses is critical for cancer genomic medicine. However, there is a significant challenge in identifying tumors with non-mutational p53 inactivation which is not detectable through DNA sequencing. These undetected cases are often misclassified as p53-normal, leading to inaccurate prognosis and downstream association analyses. To address this issue, we built the support vector machine (SVM) models to systematically reassess p53’s functional statuses in TP53 wild-type (TP53WT) tumors from multiple The Cancer Genome Atlas (TCGA) cohorts. Cross-validation demonstrated the good performance of the SVM models with a mean area under the receiver operating characteristic curve (AUROC) of 0.9822, precision of 0.9747, and recall of 0.9784. Our study revealed that a significant proportion (87%–99%) of TP53WT tumors actually had compromised p53 function. Additional analyses uncovered that these genetically intact but functionally impaired (termed as predictively reduced function of p53 or TP53WT-pRF) tumors exhibited genomic and pathophysiologic features akin to TP53-mutant tumors: heightened genomic instability and elevated levels of hypoxia. Clinically, patients with TP53WT-pRF tumors experienced significantly shortened overall survival or progression-free survival compared to those with predictively normal function of p53 (TP53WT-pN) tumors, and these patients also displayed increased sensitivity to platinum-based chemotherapy and radiation therapy.

Abstract Current studies regarding the secondary use of electronic health records (EHR) predominantly rely on domain expertise and existing medical knowledge. Though significant efforts have been devoted to investigating the application of machine learning algorithms in the EHR, efficient and powerful representation of patients is needed to unleash the potential of discovering new medical patterns underlying the EHR. Here, we present an unsupervised method for embedding high-dimensional EHR data at the patient level, aimed at characterizing patient heterogeneity in complex diseases and identifying new disease patterns associated with clinical outcome disparities. Inspired by the architecture of modern language models—specifically transformers with attention mechanisms, we use patient diagnosis and procedure codes as vocabularies and treat each patient as a sentence to perform the patient embedding. We applied this approach to 34,851 unique medical codes across 1,046,649 longitudinal patient events, including 102,739 patients from the electronic Medical Records and GEnomics (eMERGE) Network. The resulting patient vectors demonstrated excellent performance in predicting future disease events (median AUROC = 0.87 within one year) and bulk phenotyping (median AUROC = 0.84). We then illustrated the utility of these patient vectors in revealing heterogeneous comorbidity patterns, exemplified by disease subtypes in colorectal cancer and systemic lupus erythematosus, and capturing distinct longitudinal disease trajectories. External validation using EHR data from the University of Washington confirmed robust model performance, with median AUROCs of 0.83 and 0.84 for bulk phenotyping tasks and disease onset prediction, respectively. Importantly, the model reproduced the clustering results of disease subtypes identified in the eMERGE cohort and uncovered variations in overall mortality among these subtypes. Together, these results underscore the potential of representation learning in EHRs to enhance patient characterization and associated clinical outcomes, thereby advancing disease forecasting and facilitating personalized medicine.

AbstractThe challenge of disease relapsed/refractory (R/R) remains a therapeutic hurdle in chimeric antigen receptor (CAR) T‐cell therapy, especially for hematological diseases, with chronic lymphocytic leukemia (CLL) being particularly resistant to CD19 CAR T cells. Currently, there is no approved CAR T‐cell therapy for CLL patients. In this study, we aimed to address this unmet medical need by choosing the B‐cell activating factor receptor (BAFF‐R) as a promising target for CAR design against CLL. BAFF‐R is essential for B‐cell survival and is consistently expressed on CLL tumors. Our research discovered that BAFF‐R CAR T‐cell therapy exerted the cytotoxic effects on both CLL cell lines and primary B cells derived from CLL patients. In addition, the CAR T cells exhibited cytotoxicity against CD19‐knockout CLL cells that are resistant to CD19 CAR T therapy. Furthermore, we were able to generate BAFF‐R CAR T cells from small blood samples collected from CLL patients and then demonstrated the cytotoxic effects of these patient‐derived CAR T cells against autologous tumor cells. Given these promising results, BAFF‐R CAR T‐cell therapy has the potential to meet the long‐standing need for an effective treatment on CLL patients.

PURPOSE Patient-centered outcomes (PCOs) are pivotal in cancer treatment, as they directly reflect patients' quality of life. Although multiple studies suggest that factors affecting breast cancer–related morbidity and survival are influenced by treatment side effects and adherence to long-term treatment, such data are generally only available on a smaller scale or from a single center. The primary challenge with collecting these data is that the outcomes are captured as free text in clinical narratives written by clinicians. MATERIALS AND METHODS Given the complexity of PCO documentation in these narratives, computerized methods are necessary to unlock the wealth of information buried in unstructured text notes that often document PCOs. Inspired by the success of large language models (LLMs), we examined the adaptability of three LLMs, GPT-2, BioGPT, and PMC-LLaMA, on PCO tasks across three institutions, Mayo Clinic, Emory University Hospital, and Stanford University. We developed an open-source framework for fine-tuning LLM that can directly extract the five different categories of PCO from the clinic notes. RESULTS We found that these LLMs without fine-tuning (zero-shot) struggle with challenging PCO extraction tasks, displaying almost random performance, even with some task-specific examples (few-shot learning). The performance of our fine-tuned, task-specific models is notably superior compared with their non–fine-tuned LLM models. Moreover, the fine-tuned GPT-2 model has demonstrated a significantly better performance than the other two larger LLMs. CONCLUSION Our discovery indicates that although LLMs serve as effective general-purpose models for tasks across various domains, they require fine-tuning when applied to the clinician domain. Our proposed approach has the potential to lead more efficient, adaptable models for PCO information extraction, reducing reliance on extensive computational resources while still delivering superior performance for specific tasks.