Banerjee, Imon

Abstract Aims The increasing need for diagnostic echocardiography tests presents challenges in preserving the quality and promptness of reports. While Large Language Models (LLMs) have proven effective in summarizing clinical texts, their application in echo remains underexplored. Methods and results Adult echocardiography studies, conducted at the Mayo Clinic from 1 January 2017 to 31 December 2017, were categorized into two groups: development (all Mayo locations except Arizona) and Arizona validation sets. We adapted open-source LLMs (Llama-2, MedAlpaca, Zephyr, and Flan-T5) using In-Context Learning and Quantized Low-Rank Adaptation fine-tuning (FT) for echo report summarization from ‘Findings’ to ‘Impressions.’ Against cardiologist-generated Impressions, the models’ performance was assessed both quantitatively with automatic metrics and qualitatively by cardiologists. The development dataset included 97 506 reports from 71 717 unique patients, predominantly male (55.4%), with an average age of 64.3 ± 15.8 years. EchoGPT, a fine-tuned Llama-2 model, outperformed other models with win rates ranging from 87% to 99% in various automatic metrics, and produced reports comparable to cardiologists in qualitative review (significantly preferred in conciseness (P < 0.001), with no significant preference in completeness, correctness, and clinical utility). Correlations between automatic and human metrics were fair to modest, with the best being RadGraph F1 scores vs. clinical utility (r = 0.42) and automatic metrics showed insensitivity (0–5% drop) to changes in measurement numbers. Conclusion EchoGPT can generate draft reports for human review and approval, helping to streamline the workflow. However, scalable evaluation approaches dedicated to echo reports remains necessary.

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.

Abstract Deep learning may detect biologically important signals embedded in tumor morphologic features that confer distinct prognoses. Tumor morphologic features were quantified to enhance patient risk stratification within DNA mismatch repair (MMR) groups using deep learning. Using a quantitative segmentation algorithm (QuantCRC) that identifies 15 distinct morphologic features, we analyzed 402 resected stage III colon carcinomas [191 deficient (d)-MMR; 189 proficient (p)-MMR] from participants in a phase III trial of FOLFOX-based adjuvant chemotherapy. Results were validated in an independent cohort (176 d-MMR; 1,094 p-MMR). Association of morphologic features with clinicopathologic variables, MMR, KRAS, BRAFV600E, and time-to-recurrence (TTR) was determined. Multivariable Cox proportional hazards models were developed to predict TTR. Tumor morphologic features differed significantly by MMR status. Cancers with p-MMR had more immature desmoplastic stroma. Tumors with d-MMR had increased inflammatory stroma, epithelial tumor-infiltrating lymphocytes (TIL), high-grade histology, mucin, and signet ring cells. Stromal subtype did not differ by BRAFV600E or KRAS status. In p-MMR tumors, multivariable analysis identified tumor-stroma ratio (TSR) as the strongest feature associated with TTR [HRadj 2.02; 95% confidence interval (CI), 1.14–3.57; P = 0.018; 3-year recurrence: 40.2% vs. 20.4%; Q1 vs. Q2–4]. Among d-MMR tumors, extent of inflammatory stroma (continuous HRadj 0.98; 95% CI, 0.96–0.99; P = 0.028; 3-year recurrence: 13.3% vs. 33.4%, Q4 vs. Q1) and N stage were the most robust prognostically. Association of TSR with TTR was independently validated. In conclusion, QuantCRC can quantify morphologic differences within MMR groups in routine tumor sections to determine their relative contributions to patient prognosis, and may elucidate relevant pathophysiologic mechanisms driving prognosis. Significance: A deep learning algorithm can quantify tumor morphologic features that may reflect underlying mechanisms driving prognosis within MMR groups. TSR was the most robust morphologic feature associated with TTR in p-MMR colon cancers. Extent of inflammatory stroma and N stage were the strongest prognostic features in d-MMR tumors. TIL density was not independently prognostic in either MMR group.

Background: Differential kidney function assessment is an important part of preoperative evaluation of various urological interventions. It is obtained through dedicated nuclear medical imaging and is not yet implemented through conventional Imaging. Objective: We assess if differential kidney function can be obtained through evaluation of contrast-enhanced computed tomography(CT) using a combination of deep learning and (2D and 3D) radiomic features. Methods: All patients who underwent kidney nuclear scanning at Mayo Clinic sites between 2018–2022 were collected. CT scans of the kidneys were obtained within a 3-month interval before or after the nuclear scans were extracted. Patients who underwent a urological or radiological intervention within this time frame were excluded. A segmentation model was used to segment both kidneys. 2D and 3D radiomics features were extracted and compared between the two kidneys to compute delta radiomics and assess its ability to predict differential kidney function. Performance was reported using receiver operating characteristics, sensitivity, and specificity. Results: Studies from Arizona & Rochester formed our internal dataset ( n = 1,159). Studies from Florida were separately processed as an external test set to validate generalizability. We obtained 323 studies from our internal sites and 39 studies from external sites. The best results were obtained by a random forest model trained on 3D delta radiomics features. This model achieved an area under curve (AUC) of 0.85 and 0.81 on internal and external test sets, while specificity and sensitivity were 0.84,0.68 on the internal set, 0.70, and 0.65 on the external set. Conclusion: This proposed automated pipeline can derive important differential kidney function information from contrast-enhanced CT and reduce the need for dedicated nuclear scans for early-stage differential kidney functional assessment. Clinical Impact: We establish a machine learning methodology for assessing differential kidney function from routine CT without the need for expensive and radioactive nuclear medicine scans.

Abstract Objective To develop a natural language processing (NLP) algorithm that can accurately extract headache frequency from free‐text clinical notes. Background Headache frequency, defined as the number of days with any headache in a month (or 4 weeks), remains a key parameter in the evaluation of treatment response to migraine preventive medications. However, due to the variations and inconsistencies in documentation by clinicians, significant challenges exist to accurately extract headache frequency from the electronic health record (EHR) by traditional NLP algorithms. Methods This was a retrospective cross‐sectional study with patients identified from two tertiary headache referral centers, Mayo Clinic Arizona and Mayo Clinic Rochester. All neurology consultation notes written by 15 specialized clinicians (11 headache specialists and 4 nurse practitioners) between 2012 and 2022 were extracted and 1915 notes were used for model fine‐tuning (90%) and testing (10%). We employed four different NLP frameworks: (1) ClinicalBERT (Bidirectional Encoder Representations from Transformers) regression model, (2) Generative Pre‐Trained Transformer‐2 (GPT‐2) Question Answering (QA) model zero‐shot, (3) GPT‐2 QA model few‐shot training fine‐tuned on clinical notes, and (4) GPT‐2 generative model few‐shot training fine‐tuned on clinical notes to generate the answer by considering the context of included text. Results The mean (standard deviation) headache frequency of our training and testing datasets were 13.4 (10.9) and 14.4 (11.2), respectively. The GPT‐2 generative model was the best‐performing model with an accuracy of 0.92 (0.91, 0.93, 95% confidence interval [CI]) and R 2 score of 0.89 (0.87, 0.90, 95% CI), and all GPT‐2–based models outperformed the ClinicalBERT model in terms of exact matching accuracy. Although the ClinicalBERT regression model had the lowest accuracy of 0.27 (0.26, 0.28), it demonstrated a high R 2 score of 0.88 (0.85, 0.89), suggesting the ClinicalBERT model can reasonably predict the headache frequency within a range of ≤ ± 3 days, and the R 2 score was higher than the GPT‐2 QA zero‐shot model or GPT‐2 QA model few‐shot training fine‐tuned model. Conclusion We developed a robust information extraction model based on a state‐of‐the‐art large language model, a GPT‐2 generative model that can extract headache frequency from EHR free‐text clinical notes with high accuracy and R 2 score. It overcame several challenges related to different ways clinicians document headache frequency that were not easily achieved by traditional NLP models. We also showed that GPT‐2–based frameworks outperformed ClinicalBERT in terms of accuracy in extracting headache frequency from clinical notes. To facilitate research in the field, we released the GPT‐2 generative model and inference code with open‐source license of community use in GitHub. Additional fine‐tuning of the algorithm might be required when applied to different health‐care systems for various clinical use cases.

Abstract Purpose: There is a need to improve current risk stratification of stage II colorectal cancer to better inform risk of recurrence and guide adjuvant chemotherapy. We sought to examine whether integration of QuantCRC, a digital pathology biomarker utilizing hematoxylin and eosin–stained slides, provides improved risk stratification over current American Society of Clinical Oncology (ASCO) guidelines. Experimental Design: ASCO and QuantCRC-integrated schemes were applied to a cohort of 398 mismatch-repair proficient (MMRP) stage II colorectal cancers from three large academic medical centers. The ASCO stage II scheme was taken from recent guidelines. The QuantCRC-integrated scheme utilized pT3 versus pT4 and a QuantCRC-derived risk classification. Evaluation of recurrence-free survival (RFS) according to these risk schemes was compared using the log-rank test and HR. Results: Integration of QuantCRC provides improved risk stratification compared with the ASCO scheme for stage II MMRP colorectal cancers. The QuantCRC-integrated scheme placed more stage II tumors in the low-risk group compared with the ASCO scheme (62.5% vs. 42.2%) without compromising excellent 3-year RFS. The QuantCRC-integrated scheme provided larger HR for both intermediate-risk (2.27; 95% CI, 1.32–3.91; P = 0.003) and high-risk (3.27; 95% CI, 1.42–7.55; P = 0.006) groups compared with ASCO intermediate-risk (1.58; 95% CI, 0.87–2.87; P = 0.1) and high-risk (2.24; 95% CI, 1.09–4.62; P = 0.03) groups. The QuantCRC-integrated risk groups remained prognostic in the subgroup of patients that did not receive any adjuvant chemotherapy. Conclusions: Incorporation of QuantCRC into risk stratification provides a powerful predictor of RFS that has potential to guide subsequent treatment and surveillance for stage II MMRP colorectal cancers.

Abstract Percutaneous Coronary Intervention (PCI) is a widely utilized standard ofcare and efficacious intervention in the management of coronary artery disease,improving prognosis and symptoms for many patients. Prediction of post-PCIoutcomes contributes to effective patient management and quality improvementinitiatives in healthcare. Our study focuses on comparative performance analysisfor state-of-the-art vision transformer (ViT) and convolutional neural network(CNN) on multimodal data analysis in a joint fusion framework. We integrateimages of electrocardiogram (ECG) data and tabular electronic health records(EHR) to predict 3 clinically relevant post-PCI (6 months) endpoints - heart failure hospitalization, all-cause mortality, and ischemic stroke. To design a comparativemodel for ViT, we proposed a new joint fusion architecture that consistsof a convolutional neural network (CNN) with a convolutional block attentionmodule (CBAM). The learned representations are combined at an intermediatelayer followed by processing these representations using a fully connectedlayer. The proposed model demonstrates excellent performance with the highestAUROC scores of 0.849, 0.913, and 0.794 for predicting heart failure hospitalization,all-cause mortality, and stroke, respectively. Surpassing the baseline EHRmodel and ViT, the proposed CNN+CBAM fusion model showcases superiorpredictive capabilities for heart failure prediction.

In addition to the traditional clinical risk factors, an increasing amount of imaging biomarkers have shown value for cardiovascular risk prediction. Clinical and imaging data are captured from a variety of data sources during multiple patient encounters and are often analyzed independently. Initial studies showed that fusion of both clinical and imaging features results in superior prognostic performance compared with traditional scores. There are different approaches to fusion modeling, combining multiple data resources to optimize predictions, each with its own advantages and disadvantages. However, manual extraction of clinical and imaging data is time and labor intensive and often not feasible in clinical practice. An automated approach for clinical and imaging data extraction is highly desirable. Convolutional neural networks and natural language processing can be utilized for the extraction of electronic medical record data, imaging studies, and free-text data. This review outlines the current status of cardiovascular risk prediction and fusion modeling; and in addition gives an overview of different artificial intelligence approaches to automatically extract data from images and electronic medical records for this purpose.