Mayo Clinic Proceedings. Digital health最新文献

Objective

To develop a machine-learning visual classification algorithm for penile diseases in order to address disparities in access to sexual health services.

Patients and Methods

We developed an image data set using original and augmented images for 5 penile diseases: herpes lesions, syphilitic chancres, balanitis, penile cancer, and genital warts. We used a U-Net architecture model for semantic pixel segmentation into background or subject image, an Inception-ResNet version 2 neural architecture to classify each pixel as diseased or nondiseased, and a salience map using GradCAM++. We trained the model on a random 91% sample of the images and evaluated the model on the remaining 9%, assessing recall (or sensitivity), precision, specificity, and F1-score. As of July 1st 2022, the model has been in use via a mobile application platform; we assessed application usage between July and October 1, 2023.

Results

Of 239 images in the validation data set, 45 (18.8%) were of genital warts, 43 (18%) were of herpes simplex virus infection (ranging from early vesicles to ulcers), 29 (12.1%) were of penile cancer, 40 (16.7%) were of balanitis, 37 (15.5%) were of syphilitic chancres, and 45 (18.8%) were nondiseased images. The overall accuracy of the model for correctly classifying images was 0.944. There were 2640 unique submissions to the mobile platform; among a random sample (n=437), 271 (62%) were from the United States, 64 (14.6%) from Singapore, 41 (9.4%) from Canada, 40 (9.2%) from the United Kingdom, and 21 (4.8%) from Vietnam.

Conclusion

We report on the development of a machine-learning model for classifying 5 penile diseases, which exhibited excellent performance.

Objective

To conduct a study comparing the performance of automated segmentation techniques using 2 different contrast-enhanced T1-weighted (CET1) magnetic resonance imaging (MRI) acquisition protocol.

Patients and Methods

We collected 100 preoperative glioblastoma (GBM) MRIs consisting of 50 IR-GRE and 50 3-dimensional fast spin echo (3D-FSE) image sets. Their gold-standard tumor segmentation mask was created based on the expert opinion of a neuroradiologist. Cases were randomly divided into training and test sets. We used the no new UNet (nnUNet) architecture pretrained on the 501-image public data set containing IR-GRE sequence image sets, followed by 2 training rounds with the IR-GRE and 3D-FSE images, respectively. For each patient, in the IR-GRE and 3D-FSE test sets, we had 2 prediction masks, one from the model fine-tuned with the IR-GRE training set and one with 3D-FSE. The dice similarity coefficients (DSCs) of the 2 sets of results for each case in the test sets were compared using the Wilcoxon tests.

Results

Models trained on 3D-FSE images outperformed IR-GRE models in lesion segmentation, with mean DSC differences of 0.057 and 0.022 in the respective test sets. For the 3D-FSE and IR-GRE test sets, the calculated P values comparing DSCs from 2 models were .02 and .61, respectively.

Conclusion

Including 3D-FSE MRI in the training data set improves segmentation performance when segmenting 3D-FSE images.

Objective

To validate deep learning models’ ability to predict post-transplantation major adverse cardiovascular events (MACE) in patients undergoing liver transplantation (LT).

Patients and Methods

We used data from Optum’s de-identified Clinformatics Data Mart Database to identify liver transplant recipients between January 2007 and March 2020. To predict post-transplantation MACE risk, we considered patients’ demographics characteristics, diagnoses, medications, and procedural data recorded back to 3 years before the LT procedure date (index date). MACE is predicted using the bidirectional gated recurrent units (BiGRU) deep learning model in different prediction interval lengths up to 5 years after the index date. In total, 18,304 liver transplant recipients (mean age, 57.4 years [SD, 12.76]; 7158 [39.1%] women) were used to develop and test the deep learning model’s performance against other baseline machine learning models. Models were optimized using 5-fold cross-validation on 80% of the cohort, and model performance was evaluated on the remaining 20% using the area under the receiver operating characteristic curve (AUC-ROC) and the area under the precision-recall curve (AUC-PR).

Results

Using different prediction intervals after the index date, the top-performing model was the deep learning model, BiGRU, and achieved an AUC-ROC of 0.841 (95% CI, 0.822-0.862) and AUC-PR of 0.578 (95% CI, 0.537-0.621) for a 30-day prediction interval after LT.

Conclusion

Using longitudinal claims data, deep learning models can efficiently predict MACE after LT, assisting clinicians in identifying high-risk candidates for further risk stratification or other management strategies to improve transplant outcomes based on important features identified by the model.

There are mounting concerns regarding inherent bias, safety, and tendency toward misinformation of large language models (LLMs), which could have significant implications in health care. This study sought to determine whether generative artificial intelligence (AI)-based simulations of physicians making life-and-death decisions in a resource-scarce environment would demonstrate bias. Thirteen questions were developed that simulated physicians treating patients in resource-limited environments. Through a random sampling of simulated physicians using OpenAI’s generative pretrained transformer (GPT-4), physicians were tasked with choosing only 1 patient to save owing to limited resources. This simulation was repeated 1000 times per question, representing 1000 unique physicians and patients each. Patients and physicians spanned a variety of demographic characteristics. All patients had similar a priori likelihood of surviving the acute illness. Overall, simulated physicians consistently demonstrated racial, gender, age, political affiliation, and sexual orientation bias in clinical decision-making. Across all demographic characteristics, physicians most frequently favored patients with similar demographic characteristics as themselves, with most pairwise comparisons showing statistical significance (P<.05). Nondescript physicians favored White, male, and young demographic characteristics. The male doctor gravitated toward the male, White, and young, whereas the female doctor typically preferred female, young, and White patients. In addition to saving patients with their own political affiliation, Democratic physicians favored Black and female patients, whereas Republicans preferred White and male demographic characteristics. Heterosexual and gay/lesbian physicians frequently saved patients of similar sexual orientation. Overall, publicly available chatbot LLMs demonstrate significant biases, which may negatively impact patient outcomes if used to support clinical care decisions without appropriate precautions.

Mobile phone applications (MPAs) for substance use disorder (SUD) treatment are increasingly used by patients. Although pilot studies have shown promising results, multiple previous systematic reviews noted insufficient evidence for MPA use in SUD treatment—many of the previously published reviews evaluated different trials. Subsequently, we aimed to conduct an umbrella review of previously published reviews investigating the efficacy of MPAs for SUD treatment, excluding nicotine/tobacco because umbrella reviews have been done in this population and the nicotine/tobacco MPA approach often differs from SUD-focused MPAs. No previous reviews have included a statistical meta-analysis of clinical trials to quantify an estimated overall effect. Seven reviews met inclusion criteria, and 17 unique studies with available data were taken from those reviews for the meta-analysis. Overall, reviews reported a lack of evidence for recommending MPAs for SUD treatment. However, MPA-delivered recovery support services, cognitive behavioral therapy, and contingency management were identified across multiple reviews as having promising evidence for SUD treatment. Hedges g effect size for an MPA reduction in substance use–related outcomes relative to the control arm was insignificant (0.137; 95% CI, −0.056 to 0.330; P=.16). In subgroup analysis, contingency management (1.29; 95% CI, 1.088-1.482; τ²=0; k=2) and cognitive behavioral therapy (0.02; 95% CI, 0.001-0.030; τ²=0; k=2) were significant. Although contingency management’s effect was large, both trials were small (samples of 40 and 30). This review includes an adapted framework for the American Psychiatric Association’s MPA guidelines that clinicians can implement to review MPAs critically with patients.

Smart speakers have gained considerable consumer adoption and research interests. Despite their innovative interaction capabilities, a notable void exists in the literature, with no comprehensive scoping review that scrutinizes and consolidates the usage of smart speakers by providers and patients. This study performed a scoping review to explore the standalone use of smart speakers in health settings, focusing on their potential to support providers and empower patients to manage their health and well-being. Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, a comprehensive search from January 2014-September 2023, using select keywords, was performed across PubMed, Web of Science, Medline, IEEE, ACM, JAMIA, Embase, CINHAL, EBSCO, and Cochrane. The literature search yielded 1546 articles, of which 59 met the inclusion criteria. The identified studies are categorized into helping patients (n=54) with themes of independent living, reducing loneliness and improving social life, aiding in patient self-care and self-management, promoting physical activity, rethinking health care and service delivery, remote patient monitoring and communication, health information queries and helping providers (n=24) with themes recording and accessing medical information, and reducing provider workload. These research studies, performed in a controlled environment with limited patients, have found smart speakers’ high feasibility, acceptability, and positive reception in patient care and support providers. Furthermore, the findings showcase opportunities to leverage and challenges to address for a future of integrating and using smart speakers seamlessly in health settings.