Matching Patients to Clinical Trials using LLaMA 2 Embeddings and Siamese Neural Network.

Patient recruitment is a key desideratum for the success of a clinical trial that entails identifying eligible patients that match the selection criteria for the trial. However, the complexity of criteria information and heterogeneity of patient data render manual analysis a burdensome and time-consuming task. In an attempt to automate patient recruitment, this work proposes a Siamese Neural Network-based model, namely Siamese-PTM. Siamese-PTM employs the pretrained LLaMA 2 model to derive contextual representations of the EHR and criteria inputs and jointly encodes them using two weight-sharing identical subnetworks. We evaluate Siamese-PTM on structured and unstructured EHR to analyze their predictive informativeness as standalone and collective feature sets. We explore a variety of deep models for Siamese-PTM's encoders and compare their performance against the Single-encoder counterparts. We develop a baseline rule-based classifier, compared to which Siamese-PTM improved performance by 40%. Furthermore, visualization of Siamese-PTM's learned embedding space reinforces its predictive robustness.

Stratifying heart failure patients with graph neural network and transformer using Electronic Health Records to optimize drug response prediction.

OBJECTIVES: Heart failure (HF) impacts millions of patients worldwide, yet the variability in treatment responses remains a major challenge for healthcare professionals. The current treatment strategies, largely derived from population based evidence, often fail to consider the unique characteristics of individual patients, resulting in suboptimal outcomes. This study aims to develop computational models that are patient-specific in predicting treatment outcomes, by utilizing a large Electronic Health Records (EHR) database. The goal is to improve drug response predictions by identifying specific HF patient subgroups that are likely to benefit from existing HF medications. MATERIALS AND METHODS: A novel, graph-based model capable of predicting treatment responses, combining Graph Neural Network and Transformer was developed. This method differs from conventional approaches by transforming a patient's EHR data into a graph structure. By defining patient subgroups based on this representation via K-Means Clustering, we were able to enhance the performance of drug response predictions. RESULTS: Leveraging EHR data from 11 627 Mayo Clinic HF patients, our model significantly outperformed traditional models in predicting drug response using NT-proBNP as a HF biomarker across five medication categories (best RMSE of 0.0043). Four distinct patient subgroups were identified with differential characteristics and outcomes, demonstrating superior predictive capabilities over existing HF subtypes (best mean RMSE of 0.0032). DISCUSSION: These results highlight the power of graph-based modeling of EHR in improving HF treatment strategies. The stratification of patients sheds light on particular patient segments that could benefit more significantly from tailored response predictions. CONCLUSIONS: Longitudinal EHR data have the potential to enhance personalized prognostic predictions through the application of graph-based AI techniques.

Synoptic Reporting by Summarizing Cancer Pathology Reports using Large Language Models.

Background: Synoptic reporting, the documenting of clinical information in a structured manner, is known to improve patient care by reducing errors, increasing readability, interoperability, and report completeness. Despite its advantages, manually synthesizing synoptic reports from narrative reports is expensive and error prone when the number of structured fields are many. While the recent revolutionary developments in Large Language Models (LLMs) have significantly advanced natural language processing, their potential for innovations in medicine is yet to be fully evaluated. Objectives: In this study, we explore the strengths and challenges of utilizing the state-of-the-art language models in the automatic synthesis of synoptic reports. Materials and Methods: We use a corpus of 7,774 cancer related, narrative pathology reports, which have annotated reference synoptic reports from Mayo Clinic EHR. Using these annotations as a reference, we reconfigure the state-of-the-art large language models, such as LLAMA-2, to generate the synoptic reports. Our annotated reference synoptic reports contain 22 unique data elements. To evaluate the accuracy of the reports generated by the LLMs, we use several metrics including the BERT F1 Score and verify our results by manual validation. Results: We show that using fine-tuned LLAMA-2 models, we can obtain BERT Score F1 of 0.86 or higher across all data elements and BERT F1 scores of 0.94 or higher on over 50% (11 of 22) of the questions. The BERT F1 scores translate to average accuracies of 76% and as high as 81% for short clinical reports. Conclusions: We demonstrate successful automatic synoptic report generation by fine-tuning large language models.

Artificial Intelligence-based Efficacy Prediction of Phase 3 Clinical Trial for Repurposing Heart Failure Therapies.

Introduction: Drug repurposing involves finding new therapeutic uses for already approved drugs, which can save costs as their pharmacokinetics and pharmacodynamics are already known. Predicting efficacy based on clinical endpoints is valuable for designing phase 3 trials and making Go/No-Go decisions, given the potential for confounding effects in phase 2. Objectives: This study aims to predict the efficacy of the repurposed Heart Failure (HF) drugs for the Phase 3 Clinical Trial. Methods: Our study presents a comprehensive framework for predicting drug efficacy in phase 3 trials, which combines drug-target prediction using biomedical knowledgebases with statistical analysis of real-world data. We developed a novel drug-target prediction model that uses low-dimensional representations of drug chemical structures and gene sequences, and biomedical knowledgebase. Furthermore, we conducted statistical analyses of electronic health records to assess the effectiveness of repurposed drugs in relation to clinical measurements (e.g., NT-proBNP). Results: We identified 24 repurposed drugs (9 with a positive effect and 15 with a non-positive) for heart failure from 266 phase 3 clinical trials. We used 25 genes related to heart failure for drug-target prediction, as well as electronic health records (EHR) from the Mayo Clinic for screening, which contained over 58,000 heart failure patients treated with various drugs and categorized by heart failure subtypes. Our proposed drug-target predictive model performed exceptionally well in all seven tests in the BETA benchmark compared to the six cutting-edge baseline methods (i.e., best performed in 266 out of 404 tasks). For the overall prediction of the 24 drugs, our model achieved an AUCROC of 82.59% and PRAUC (average precision) of 73.39%. Conclusion: The study demonstrated exceptional results in predicting the efficacy of repurposed drugs for phase 3 clinical trials, highlighting the potential of this method to facilitate computational drug repurposing.

Using artificial intelligence to learn optimal regimen plan for Alzheimer's disease.

BACKGROUND: Alzheimer's disease (AD) is a progressive neurological disorder with no specific curative medications. Sophisticated clinical skills are crucial to optimize treatment regimens given the multiple coexisting comorbidities in the patient population. OBJECTIVE: Here, we propose a study to leverage reinforcement learning (RL) to learn the clinicians' decisions for AD patients based on the longitude data from electronic health records. METHODS: In this study, we selected 1736 patients from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. We focused on the two most frequent concomitant diseases-depression, and hypertension, thus creating 5 data cohorts (ie, Whole Data, AD, AD-Hypertension, AD-Depression, and AD-Depression-Hypertension). We modeled the treatment learning into an RL problem by defining states, actions, and rewards. We built a regression model and decision tree to generate multiple states, used six combinations of medications (ie, cholinesterase inhibitors, memantine, memantine-cholinesterase inhibitors, hypertension drugs, supplements, or no drugs) as actions, and Mini-Mental State Exam (MMSE) scores as rewards. RESULTS: Given the proper dataset, the RL model can generate an optimal policy (regimen plan) that outperforms the clinician's treatment regimen. Optimal policies (ie, policy iteration and Q-learning) had lower rewards than the clinician's policy (mean -3.03 and -2.93 vs. -2.93, respectively) for smaller datasets but had higher rewards for larger datasets (mean -4.68 and -2.82 vs. -4.57, respectively). CONCLUSIONS: Our results highlight the potential of using RL to generate the optimal treatment based on the patients' longitude records. Our work can lead the path towards developing RL-based decision support systems that could help manage AD with comorbidities.

Leveraging multi-source to resolve inconsistency across pharmacogenomic datasets in drug sensitivity prediction.

Pharmacogenomics datasets have been generated for various purposes, such as investigating different biomarkers. However, when studying the same cell line with the same drugs, differences in drug responses exist between studies. These variations arise from factors such as inter-tumoral heterogeneity, experimental standardization, and the complexity of cell subtypes. Consequently, drug response prediction suffers from limited generalizability. To address these challenges, we propose a computational model based on Federated Learning (FL) for drug response prediction. By leveraging three pharmacogenomics datasets (CCLE, GDSC2, and gCSI), we evaluate the performance of our model across diverse cell line-based databases. Our results demonstrate superior predictive performance compared to baseline methods and traditional FL approaches through various experimental tests. This study underscores the potential of employing FL to leverage multiple data sources, enabling the development of generalized models that account for inconsistencies among pharmacogenomics datasets. By addressing the limitations of low generalizability, our approach contributes to advancing drug response prediction in precision oncology.

Phosphorylation alters the mechanical stiffness of a model fragment of the dystrophin homologue utrophin.

Duchenne muscular dystrophy is a lethal muscle wasting disease caused by the absence of the protein dystrophin. Utrophin is a dystrophin homologue currently under investigation as a protein replacement therapy for Duchenne muscular dystrophy. Dystrophin is hypothesized to function as a molecular shock absorber that mechanically stabilizes the sarcolemma. While utrophin is homologous with dystrophin from a molecular and biochemical perspective, we have recently shown that full-length utrophin expressed in eukaryotic cells is stiffer than what has been reported for dystrophin fragments expressed in bacteria. In this study, we show that differences in expression system impact the mechanical stiffness of a model utrophin fragment encoding the N terminus through spectrin repeat 3 (UtrN-R3). We also demonstrate that UtrN-R3 expressed in eukaryotic cells was phosphorylated while bacterial UtrN-R3 was not detectably phosphorylated. Using atomic force microscopy, we show that phosphorylated UtrN-R3 exhibited significantly higher unfolding forces compared to unphosphorylated UtrN-R3 without altering its actin-binding activity. Consistent with the effect of phosphorylation on mechanical stiffness, mutating the phosphorylated serine residues on insect eukaryotic protein to alanine decreased its stiffness to levels not different from unphosphorylated bacterial protein. Taken together, our data suggest that the mechanical properties of utrophin may be tuned by phosphorylation, with the potential to improve its efficacy as a protein replacement therapy for dystrophinopathies.

Distinct mechanical properties in homologous spectrin-like repeats of utrophin.

Patients with Duchenne muscular dystrophy (DMD) lack the protein dystrophin, which is a critical molecular component of the dystrophin-glycoprotein complex (DGC). Dystrophin is hypothesized to function as a molecular shock absorber that mechanically stabilizes the sarcolemma of striated muscle through interaction with the cortical actin cytoskeleton via its N-terminal half and with the transmembrane protein ß-dystroglycan via its C-terminal region. Utrophin is a fetal homologue of dystrophin that can subserve many dystrophin functions and is therefore under active investigation as a dystrophin replacement therapy for DMD. Here, we report the first mechanical characterization of utrophin using atomic force microscopy (AFM). Our data indicate that the mechanical properties of spectrin-like repeats in utrophin are more in line with the PEVK and Ig-like repeats of titin rather than those reported for repeats in spectrin or dystrophin. Moreover, we measured markedly different unfolding characteristics for spectrin repeats within the N-terminal actin-binding half of utrophin compared to those in the C-terminal dystroglycan-binding half, even though they exhibit identical thermal denaturation profiles. Our results demonstrate dramatic differences in the mechanical properties of structurally homologous utrophin constructs and suggest that utrophin may function as a stiff elastic element in series with titin at the myotendinous junction.