News & Comment

The integration of Quality Management System (QMS) principles into the life cycle of development, deployment, and utilization of machine learning (ML) and artificial intelligence (AI) technologies within healthcare settings holds the potential to close the AI translation gap by establishing a robust framework that accelerates the safe, ethical, and effective delivery of AI/ML in day-to-day patient care. Healthcare organizations (HCOs) can implement these principles effectively by embracing an enterprise QMS analogous to those in regulated industries. By establishing a QMS explicitly tailored to health AI technologies, HCOs can comply with evolving regulations and minimize redundancy and rework while aligning their internal governance practices with their steadfast commitment to scientific rigor and medical excellence.

As result of the pandemic-related increase in telehealth and the 21st Century Cures Act, technology is playing an increasing role in healthcare. This has led to organizational investments in the “digital front door” of healthcare. The promise that these technologies can revolutionize care by better connecting us to our patients, overcoming analog barriers to care, and addressing health disparities is grounded in “techno-optimism.” We advocate for organizational leaders to inform their digital health equity strategies with a balanced measure of “techno-skepticism”, grounded in implementation science, that can ensure alignment between health technology and health equity.

At a time of diminishing clinical skills and diagnostic endeavours for clinicians worldwide, a method of rapid automated microscopy coupled with pH measurement is introduced with early clinical success results in women with common vulvovaginal symptoms facilitating rapid diagnosis and enhanced therapeutic measures.

Sleep recordings are visually classified in stages by experts in the field, based on consensus international criteria. This procedure is expensive and time-consuming. Automatic sleep scoring systems have, progressively over the years, demonstrated good levels of accuracy. Although the performance of these algorithms is believed to be high, however, there remains widespread skepticism in their daily use in clinical and scientific practice. In this comment to a recent publication of NPJ Digital Medicine, we express the reasons why we think the sleep expert should remain the central pivot in the pendulum between visual and automatic methodology, trying to find a new balance in the scientific debate.

Digital health technologies (DHTs) enable remote data collection, support a patient-centric approach to drug development, and provide real-time data in real-world settings. With increasing use of DHTs in clinical care and development, we expect a growing body of evidence supporting use of DHTs to capture endpoint data in clinical trials. As the body of evidence grows, it will be critical to ensure that available prior work can be leveraged. We propose a framework to reuse analytical and clinical validation, as well as verification data, generated for existing DHTs. We apply real life case studies to illustrate our proposal aimed at leveraging prior work, while applying the V3 framework (verification, analytical validation, clinical validation) and avoiding duplication. Utilizing our framework will enable stakeholders to share best practices and consistent approaches to employing these tools in clinical studies, build on each other’s work, and ultimately accelerate evidence generation demonstrating the reproducibility and value add of these new tools.

Artificial Intelligence-supported digital applications (AI applications) are expected to transform radiology. However, providers need the motivation and incentives to adopt these technologies. For some radiology AI applications, the benefits of the application itself may sufficiently serve as the incentive. For others, payers may have to consider reimbursing the AI application separate from the cost of the underlying imaging studies. In such circumstances, it is important for payers to develop a clear set of criteria to decide which AI applications should be paid for separately. In this article, we propose a framework to help serve as a guide for payers aiming to establish such criteria and for technology vendors developing radiology AI applications. As a rule of thumb, we propose that radiology AI applications with a clinical utility must be reimbursed separately provided they have supporting evidence that the improved diagnostic performance leads to improved outcomes from a societal standpoint, or if such improved outcomes can reasonably be anticipated based on the clinical utility offered.

The generalizability of predictive algorithms is of key relevance to application in clinical practice. We provide an overview of three types of generalizability, based on existing literature: temporal, geographical, and domain generalizability. These generalizability types are linked to their associated goals, methodology, and stakeholders.

Digital tools are transforming mental health care. The promise of this transformation to improve outcomes has not yet been realized fully. While some have become skeptical, this article argues that we are just at the end of Act 1, with several opportunities and challenges ahead.

Postpartum mental health conditions are a public health concern, affecting a large number of reproductive-age women and their families. Postpartum depression alone affects at least 14% of new mothers and their families. However, very little has been written about how advances in digital mental health can benefit women in the postpartum period, or how those advances may poorly serve this vulnerable population. This manuscript takes a high-level view of the advances in different areas of digital mental health, including telehealth, apps, and digital phenotyping. In this comment, we explore ways in which digital interventions for postpartum mental health may help with connection to treatment, accessibility, agency, and ease of access. We also note particular concerns for how digital postpartum mental health may encounter issues of low-quality resources, ethical considerations, and equity considerations. We provide suggestions for how to leverage the promise and avoid the pitfalls of digital mental health for postpartum women.

This paper reviews the current state of patient safety and the application of artificial intelligence (AI) techniques to patient safety. This paper defines patient safety broadly, not just inpatient care but across the continuum of care, including diagnostic errors, misdiagnosis, adverse events, injuries, and measurement issues. It outlines the major current uses of AI in patient safety and the relative adoption of these techniques in hospitals and health systems. It also outlines some of the limitations of these AI systems and the challenges with evaluation of these systems. Finally, it outlines the importance of developing a proactive agenda for AI in healthcare that includes marked increased funding of research and evaluation in this area.