Breast cancer is a pervasive global health concern among women. Leveraging multimodal data from enterprise patient databases-including Picture Archiving and Communication Systems (PACS) and Electronic Health Records (EHRs)-holds promise for improving prediction. This study introduces a multimodal deep-learning model leveraging mammogram datasets to evaluate breast cancer prediction. Our approach integrates frozen large-scale pretrained vision-language models, showcasing superior performance and stability compared to traditional image-tabular models across two public breast cancer datasets. The model consistently outperforms conventional full fine-tuning methods by using frozen pretrained vision-language models alongside a lightweight trainable classifier. The observed improvements are significant. In the CBIS-DDSM dataset, the Area Under the Curve (AUC) increases from 0.867 to 0.902 during validation and from 0.803 to 0.830 for the official test set. Within the EMBED dataset, AUC improves from 0.780 to 0.805 during validation. In scenarios with limited data, using Breast Imaging-Reporting and Data System category three (BI-RADS 3) cases, AUC improves from 0.91 to 0.96 on the official CBIS-DDSM test set and from 0.79 to 0.83 on a challenging validation set. This study underscores the benefits of vision-language models in jointly training diverse image-clinical datasets from multiple healthcare institutions, effectively addressing challenges related to non-aligned tabular features. Combining training data enhances breast cancer prediction on the EMBED dataset, outperforming all other experiments. In summary, our research emphasizes the efficacy of frozen large-scale pretrained vision-language models in multimodal breast cancer prediction, offering superior performance and stability over conventional methods, reinforcing their potential for breast cancer prediction.

This paper presents a novel Lasso Logistic Regression model based on feature-based time series data to determine disease severity and when to administer drugs or escalate intervention procedures in patients with coronavirus disease 2019 (COVID-19). Advanced features were extracted from highly enriched and time series vital sign data of hospitalized COVID-19 patients, including oxygen saturation readings, and with a combination of patient demographic and comorbidity information, as inputs into the dynamic feature-based classification model. Such dynamic combinations brought deep insights to guide clinical decision-making of complex COVID-19 cases, including prognosis prediction, timing of drug administration, admission to intensive care units, and application of intervention procedures like ventilation and intubation. The COVID-19 patient classification model was developed utilizing 900 hospitalized COVID-19 patients in a leading multi-hospital system in Texas, United States. By providing mortality prediction based on time-series physiologic data, demographics, and clinical records of individual COVID-19 patients, the dynamic feature-based classification model can be used to improve efficacy of the COVID-19 patient treatment, prioritize medical resources, and reduce casualties. The uniqueness of our model is that it is based on just the first 24 hours of vital sign data such that clinical interventions can be decided early and applied effectively. Such a strategy could be extended to prioritize resource allocations and drug treatment for futurepandemic events.

To evaluate the performance of a biopsy decision support algorithmic model, the intelligent-augmented breast cancer risk calculator (iBRISK), on a multicenter patient dataset. iBRISK was previously developed by applying deep learning to clinical risk factors and mammographic descriptors from 9700 patient records at the primary institution and validated using another 1078 patients. All patients were seen from March 2006 to December 2016. In this multicenter study, iBRISK was further assessed on an independent, retrospective dataset (January 2015–June 2019) from three major health care institutions in Texas, with Breast Imaging Reporting and Data System (BI-RADS) category 4 lesions. Data were dichotomized and trichotomized to measure precision in risk stratification and probability of malignancy (POM) estimation. iBRISK score was also evaluated as a continuous predictor of malignancy, and cost savings analysis was performed. The iBRISK model's accuracy was 89.5%, area under the receiver operating characteristic curve (AUC) was 0.93 (95% CI: 0.92, 0.95), sensitivity was 100%, and specificity was 81%. A total of 4209 women (median age, 56 years [IQR, 45–65 years]) were included in the multicenter dataset. Only two of 1228 patients (0.16%) in the “low” POM group had malignant lesions, while in the “high” POM group, the malignancy rate was 85.9%. iBRISK score as a continuous predictor of malignancy yielded an AUC of 0.97 (95% CI: 0.97, 0.98). Estimated potential cost savings were more than $420 million. iBRISK demonstrated high sensitivity in the malignancy prediction of BI-RADS 4 lesions. iBRISK may safely obviate biopsies in up to 50% of patients in low or moderate POM groups and reduce biopsy-associated costs. Keywords: Mammography, Breast, Oncology, Biopsy/Needle Aspiration, Radiomics, Precision Mammography, AI-augmented Biopsy Decision Support Tool, Breast Cancer Risk Calculator, BI-RADS 4 Mammography Risk Stratification, Overbiopsy Reduction, Probability of Malignancy (POM) Assessment, Biopsy-based Positive Predictive Value (PPV3) Supplemental material is available for this article. Published under a CC BY 4.0 license. See also the commentary by McDonald and Conant in this issue.

Personalized genomics instability, e.g., somatic mutations, is believed to contribute to the heterogeneous drug responses in patient cohorts. However, it is difficult to discover personalized driver mutations that are predictive of drug sensitivity owing to diverse and complex mutations of individual patients. To circumvent this problem, a novel computational method is presented to discover potential drug sensitivity relevant cancer subtypes and identify driver mutation modules of individual subtypes by coupling differentially expressed genes (DEGs) based subtyping analysis with the driver mutation network analysis.

Probability of malignancy for BI-RADS 4-designated breast lesions ranges from 2% to 95%, contributing to high false-positive biopsy rates. We compare clinical performance of digital breast tomosynthesis (DBT) versus digital mammography (2D) among our BI-RADS 4 population without prior history of breast cancer. We extracted retrospective data i.e., clinical, mammogram reports, and biopsy data, from electronic medical records across Houston Methodist’s nine hospitals for patients who underwent diagnostic examinations using both modalities (02/01/2015 – 09/30/2020). 2D and DBT cohorts were not intra-individual matched, and there was no direct mammogram evaluation. Using Student’s t test, Fisher’s exact test, and Chi-squared test, we evaluated the data to determine statistical significance of differences between modalities in BI-RADS 4 cases. We calculated adjusted odds-ratio between modalities for cancer detection rate (CDR) and biopsy-derived positive predictive value (PPV3). There were 6,356 encounters (6,020 patients) in 2D and 5,896 encounters (5,637 patients) in DBT assessed as BI-RADS 4. Using Fisher’s exact test, DBT mammography cases were significantly assessed as BI-RADS 4 5.66% more often than those undergoing 2D mammography, P = 0.0046 (1.0566 95% CI: 1.0169–1.0977). The CDRs were 112.65 (2D) and 120.76 (DBT), adjusted odds-ratio: 1.04 (0.93, 1.16)), P = 0.5029, while PPV3 were 14.41% (2D) and 15.99% (DBT), adjusted odds-ratio: 1.09 (0.97, 1.22), P = 0.1483; both logistic regression-adjusted for all other factors. DBT did not achieve better performance and sensitivity in assigning BI-RADS 4 cases compared with 2D, showed no significant advantage in CDR and PPV3, and does not reduce false-positive biopsies among BI-RADS 4-assessed patients.

Patients with inflammatory bowel disease often present to the emergency department due to the chronic relapsing nature of the disease. Previous studies have shown younger patients to have an increased frequency of emergency department visits, resulting in repeated exposure to imaging studies and steroids, both of which are associated with risks. We performed a retrospective cohort analysis of inflammatory bowel disease patients seen at Houston Methodist Hospital’s emergency department from January 2014 to December 2017 using ICD codes to identify patients with Crohn’s disease, ulcerative colitis, or indeterminate colitis from the electronic medical record. Data were collected on demographics, medications, and imaging. Five hundred and fifty-nine patients were randomly selected for inclusion. Older age was associated with decreased risk of CT scan or steroid use. Patients with ulcerative colitis compared to Crohn’s had decreased risk of CT scan, while there was an increased risk of CT in patients on a biologic, immunomodulator, or when steroids were given. Steroid use was also more common in those with inflammatory bowel disease as the primary reason for the visit. Patients in our study frequently received steroids and had CT scans performed. The increased risk of CT in those on a biologic, immunomodulator, or steroids suggests more severe disease may contribute. Guidelines are needed to reduce any unnecessary corticosteroid use and limit repeat CT scans in young inflammatory bowel disease patients to decrease the risk of radiation-associated malignancy over their lifetime.