Infections in Diabetic Foot Ulcers (DFUs) can cause severe complications, including tissue death and limb amputation, highlighting the need for accurate, timely diagnosis. Previous machine learning methods have focused on identifying infections by analyzing wound images alone, without utilizing additional metadata such as medical notes. In this study, we aim to improve infection classification by introducing Synthetic Caption Augmented Retrieval for Wound Infection Detection (SCARWID) , a novel multimodal deep learning framework that leverages synthetic textual descriptions to augment DFU images. SCARWID consists of two components: (1) Wound-BLIP, a Vision-Language Model (VLM) fine-tuned on GPT-4o-generated descriptions to synthesize consistent captions from images; and (2) an Image–Text Fusion module that uses cross-attention to extract cross-modal embeddings from an image and its corresponding Wound-BLIP caption. Infection status is determined by retrieving the top- \( k \) similar items from a labeled support set. To enhance the diversity of training data, we utilized a latent diffusion model to generate additional wound images. As a result, SCARWID outperformed state-of-the-art models, achieving average accuracy, sensitivity, and specificity of 0.814, 0.845, and 0.783, respectively, for wound infection classification.

Automatic assessment of wounds using smartphones can facilitate the tracking of healing progress outside the clinic. Prior work proposed SS-PMG-EfficientNets, a state-of-the-art, multi-scale, convolutional neural network architecture for accurate wound assessment. However, the model was previously only trained and evaluated on high-quality wound images captured in controlled environments. This study systematically evaluated the effects of low-quality images on SS-PMG-EfficientNets, finding that the model's performance and consistency are reduced when input wound images are blurry or captured in adverse lighting (too bright or too dark), two factors previously identified as the most prevalent issues in crowdsourced images. Specifically, SS-PMG-EfficientNet was first trained on three high-quality wound datasets containing images labeled with Photographic Wound Assessment Tool (PWAT) wound healing scores, stages of pressure injury wounds, and wound types, respectively. The model was then evaluated on a test wound dataset that was augmented with real-world image luminance, blur, and image compression effects and found that wound assessment accuracy decreased by up to 27%. To improve model performance, we propose a method for instant image quality assessment on the smartphone, enabling low-quality wound images to be rejected immediately after image capture so that the nurse or patient can be prompted to recapture the image. Our evaluation of the proposed image quality assessment and rejection method showed that it increases average accuracy from 73 % to 87%.

OBJECTIVE: We sought to understand how patients and primary care teams use secure messaging (SM) to communicate with one another by analyzing secure message threads from 2 Department of Veterans Affairs facilities. METHODS: We coded 1000 threads of SM communication sampled from 40 primary care teams. RESULTS: Most threads (94.5%) were initiated by patients (90.4%) or caregivers (4.1%); only 5.5% were initiated by primary care team members proactively reaching out to patients. Medication renewals and refills (47.2%), scheduling requests (17.6%), medication issues (12.9%), and health issues (12.7%) were the most common patient-initiated requests, followed by referrals (7.0%), administrative issues (6.5%), test results (5.4%), test issues (5.2%), informing messages (4.9%), comments about the patient portal or SM (4.1%), appreciation (3.9%), self-reported data (2.8%), life issues (1.5%), and complaints (1.5%). Very few messages were clinically urgent (0.7%) or contained other potentially challenging content. Message threads were mostly short (2.7 messages), comprising an average of 1.35 discrete content types. A substantial proportion of issues (24.2%) did not show any evidence of being resolved through SM. Time to response and extent of resolution via SM varied by message content. Proactive SM use by teams varied, but was most often for test results (32.7%), medication-related issues (21.8%), medication renewals (16.4%), or scheduling issues (18.2%). CONCLUSIONS: The majority of messages were transactional and initiated by patients or caregivers. Not all content categories were fully addressed over SM. Further education and training for both patients and clinical teams could improve the quality and efficiency of SM communication.

Over one billion smartphones have now been shipped worldwide. These mobile devices feature multi-core CPUs and GPUs, megapixel cameras and an array of sensors. Smartphone sensors can now be processed to diagnose a wide variety of medical conditions including cough detection, irregular heartbeat detection, and lung function analysis. The ability to diagnose ailments in the convenience of patients' homes on smartphones that they already possess could lead to early detection, which could ultimately reduce healthcare costs. This paper reviews state-of-the-art examples, examines the technical issues involved in the use of the smartphone as a medical device, and outlines potential benefits and challenges. A case study of an Android smartphone app for wound detection and healing progress analysis by diabetes patients is also presented.

The active engagement of consumers is an important factor in achieving widespread success of health information systems. The disability community represents a major segment of the healthcare arena, with more than 50 million Americans experiencing some form of disability. In keeping with the "consumer-driven" approach to e-health systems, this paper considers the distinctive aspects of electronic and personal health record use by this segment of society. Drawing upon the information shared during two national policy forums on this topic, the authors present the concept of Electronic Disability Records (EDR). The authors outline the purpose and parameters of such records, with specific attention to its ability to organize health and financial data in a manner that can be used to expedite the disability determination process. In doing so, the authors discuss its interaction with Electronic Health Records (EHR) and Personal Health Records (PHR). The authors then draw upon these general parameters to outline a model use case for disability determination and discuss related implications for disability health management. The paper further reports on the subsequent considerations of these and related deliberations by the American Health Information Community (AHIC).

The next generation of enterprise networks is undergoing major changes as a plethora of new architectures, applications, and services begin to roll out within businesses. In general, the world of voice/telephony, video, and data are "converging" into a global communications network. The purpose of this paper is twofold. First, the design, analysis, and performance of a session initiation protocol (SIP)-based videoconferencing desktop client, which has been developed and deployed over Internet2, is presented. Second, a guideline for managing SIP-based services to be deployed within enterprises, which addresses several challenges in each layer, such as network address translator (NAT)/FW issues, directory service integration issues, and interoperability issues, is proposed. Several detailed experimental results related to interoperability and conformance that were carried out are presented. Findings of extensive SIP/NAT traversal analysis through network traffic measurements are reported. The lessons learned from both the design of a new SIP-based voice/video client, as well as management challenges with enterprise deployment are highlighted.