AI in Pathology: What could possibly go wrong?

The field of medicine is undergoing rapid digital transformation. Pathologists are now striving to digitize their data, workflows, and interpretations, assisted by the enabling development of whole-slide imaging. Going digital means that the analog process of human diagnosis can be augmented or even replaced by rapidly evolving AI approaches, which are just now entering into clinical practice. But with such progress comes challenges that reflect a variety of stressors, including the impact of unrepresentative training data with accompanying implicit bias, data privacy concerns, and fragility of algorithm performance. Beyond such core digital aspects, considerations arise related to difficulties presented by changing disease presentations, diagnostic approaches, and therapeutic options. While some tools such as data federation can help with broadening data diversity while preserving expertise and local control, they may not be the full answer to some of these issues. The impact of AI in pathology on the field's human practitioners is still very much unknown: installation of unconscious bias and deference to AI guidance need to be understood and addressed. If AI is widely adopted, it may remove many inefficiencies in daily practice and compensate for staff shortages. It may also cause practitioner deskilling, dethrilling, and burnout. We discuss the technological, clinical, legal, and sociological factors that will influence the adoption of AI in pathology, and its eventual impact for good or ill.

Closing the Referral Loop: an Analysis of Primary Care Referrals to Specialists in a Large Health System.

PURPOSE: Ideally, a referral from a primary care physician (PCP) to a specialist results in a completed specialty appointment with results available to the PCP. This is defined as "closing the referral loop." As health systems grow more complex, regulatory bodies increase vigilance, and reimbursement shifts towards value, closing the referral loop becomes a patient safety, regulatory, and financial imperative. OBJECTIVE/DESIGN: To assess the ability of a large health system to close the referral loop, we used electronic medical record (EMR)-generated data to analyze referrals from a large primary care network to 20 high-volume specialties between July 1, 2015 and June 30, 2016. MAIN MEASURES: The primary metric was documented specialist appointment completion rate. Explanatory analyses included documented appointment scheduling rate, individual clinic differences, appointment wait times, and geographic distance to appointments. KEY RESULTS: Of the 103,737 analyzed referral scheduling attempts, only 36,072 (34.8%) resulted in documented complete appointments. Low documented appointment scheduling rates (38.9% of scheduling attempts lacked appointment dates), individual clinic differences in closing the referral loop, and significant differences in wait times and distances to specialists between complete and incomplete appointments drove this gap. Other notable findings include high variation in wait times among specialties and correlation between high wait times and low documented appointment completion rates. CONCLUSIONS: The rate of closing the referral loop in this health system is low. Low appointment scheduling rates, individual clinic differences, and patient access issues of wait times and geographic proximity explain much of the gap. This problem is likely common among large health systems with complex provider networks and referral scheduling. Strategies that improve scheduling, decrease variation among clinics, and improve patient access will likely improve rates of closing the referral loop. More research is necessary to determine the impact of these changes and other potential driving factors.

CAFES: Chest X-ray Analysis using Federated Self-supervised Learning for Pediatric COVID-19 Detection.

Chest X-rays (CXRs) play a pivotal role in cost-effective clinical assessment of various heart and lung related conditions. The urgency of COVID-19 diagnosis prompted their use in identifying conditions like lung opacity, pneumonia, and acute respiratory distress syndrome in pediatric patients. We propose an AI-driven solution for binary COVID-19 versus non-COVID-19 classification in pediatric CXRs. We present a Federated Self-Supervised Learning (FSSL) framework to enhance Vision Transformer (ViT) performance for COVID-19 detection in pediatric CXRs. ViT's prowess in vision-related binary classification tasks, combined with self-supervised pre-training on adult CXR data, forms the basis of the FSSL approach. We implement our strategy on the Rhino Health Federated Computing Platform (FCP), which ensures privacy and scalability for distributed data. The chest X-ray analysis using the federated SSL (CAFES) model, utilizes the FSSL-pre-trained ViT weights and demonstrated gains in accurately detecting COVID-19 when compared with a fully supervised model. Our FSSL-pre-trained ViT showed an area under the precision-recall curve (AUPR) of 0.952, which is 0.231 points higher than the fully supervised model for COVID-19 diagnosis using pediatric data. Our contributions include leveraging vision transformers for effective COVID-19 diagnosis from pediatric CXRs, employing distributed federated learning-based self-supervised pre-training on adult data, and improving pediatric COVID-19 diagnosis performance. This privacy-conscious approach aligns with HIPAA guidelines, paving the way for broader medical imaging applications.

Accelerating artificial intelligence: How federated learning can protect privacy, facilitate collaboration, and improve outcomes.

Cross-institution collaborations are constrained by data-sharing challenges. These challenges hamper innovation, particularly in artificial intelligence, where models require diverse data to ensure strong performance. Federated learning (FL) solves data-sharing challenges. In typical collaborations, data is sent to a central repository where models are trained. With FL, models are sent to participating sites, trained locally, and model weights aggregated to create a master model with improved performance. At the 2021 Radiology Society of North America's (RSNA) conference, a panel was conducted titled "Accelerating AI: How Federated Learning Can Protect Privacy, Facilitate Collaboration and Improve Outcomes." Two groups shared insights: researchers from the EXAM study (EMC CXR AI Model) and members of the National Cancer Institute's Early Detection Research Network's (EDRN) pancreatic cancer working group. EXAM brought together 20 institutions to create a model to predict oxygen requirements of patients seen in the emergency department with COVID-19 symptoms. The EDRN collaboration is focused on improving outcomes for pancreatic cancer patients through earlier detection. This paper describes major insights from the panel, including direct quotes. The panelists described the impetus for FL, the long-term potential vision of FL, challenges faced in FL, and the immediate path forward for FL.

Keeping care connected: e-Consultation program improves access to nephrology care.

INTRODUCTION: Health systems are seeking innovative solutions to improve specialty care access. Electronic consultations (eConsults) allow specialists to provide formal clinical recommendations to primary care providers (PCPs) based on patient chart review, without a face-to-face visit. METHODS: We implemented a nephrology eConsult pilot program within a large, academic primary care practice to facilitate timely communication between nephrologists and PCPs. We used primary care referral data to compare wait times and completion rates between traditional referrals and eConsults. We surveyed PCPs to assess satisfaction with the program. RESULTS: For traditional nephrology referrals placed during the study period (July 2016-March 2017), there was a 51-day median appointment wait time and a 40.9% referral completion rate. For eConsults, there was a median nephrologist response time of one day and a 100% completion rate; 67.5% of eConsults did not require a subsequent face-to-face specialty appointment. For eConsults that were converted to an in-person visit, the median wait time and completion rate were 40 days and 73.1%, respectively. Compared to traditional referrals placed during the study period, eConsults converted to in-person visits were more likely to be completed ( p = 0.001). Survey responses revealed that PCPs were highly satisfied with the program and consider the quick turnaround time as the greatest benefit. DISCUSSION: Our eConsult pilot program reduced nephrology wait times and significantly increased referral completion rates. In large integrated health systems, eConsults have considerable potential to improve access to specialty care, reduce unnecessary appointments, and optimize the patient population being seen by specialists.

Engaging the primary care community to encourage appropriate prostate cancer screening.

Prostate-specific antigen (PSA) screening for prostate cancer remains a controversial topic, particularly in the primary care community. Our multidisciplinary prostate screening panel at Duke University Health System, USA created a nuanced PSA screening algorithm, implemented it into the Electronic Health Record of Duke Primary Care, and conducted outreach meetings with primary care practices to support its rollout. Through this project, we identified areas of concern among primary care clinicians regarding PSA screening that we structured into two major categories: ideological opposition and logistical opposition. We outlined specific concerns in each major category and described how our team responded to those concerns. As communication between primary care clinicians and prostate specialists is vital to the success and safety of PSA screening programs, we hope that describing primary care concerns and our responses to them will help other health systems thoughtfully and efficiently implement appropriate PSA screening programs moving forward.