How using AI in pathology can aid healthcare equity and efficiency

Wider use of advanced computing in pathology can help organizations achieve important aspects of the Quintuple Aim.

This article is part of AI BEYOND the Hype - March/April 2024 COVERstory.

The application of artificial intelligence to pathology continues to advance and offers the compelling promise of addressing all aspects of the quintuple aim.

In light of the centrality of pathology in the diagnosis, treatment and monitoring of patients in many disease areas, the adoption of digital pathology and AI can have lasting positive impact on patient outcomes, physician experience and the economy at large.

Better outcomes

Improvement in population health outcomes is enabled by robust research and development efforts that give rise to the best possible treatment for each patient with a given disease. The application of AI in pathology is transforming translational research and clinical trials in many ways.

In translational research, AI is enabling the aggregation and analysis of big data — from pathology, radiology, molecular, clinical, wearables — to derive new insights on biomarkers and disease pathways at an impressive scale. In the assessment of solid tumors, AI is enabling precise quantification of biomarkers that are relevant to clinical outcomes.

Additionally, the application of AI is providing insights into the spatial complexity of the tumor microenvironment. New insights gained from the troves of data hold the promise to unravel disease pathways giving rise to new treatment targets and more nuanced biomarkers to predict individual patient responses to treatment. Some of these new insights previously could not be extracted let alone analyzed.

In clinical trials, AI enables selection of highly specific patient populations that are most likely to respond to therapy, while accurately and reproducibly measuring individual patient response based on biomarkers that are challenging to assess manually. One example is metabolic dysfunction-associated steatohepatitis (MASH), where machine learning enables measurement of therapy-relevant attributes on a continuous scale that eludes manual approaches.

In MASH, drug efficacy based on histological endpoints is required to initiate phase 3 trials, and the subjectivity of manual pathology-based endpoint scoring can negatively impact the accuracy of patient stratification and the likelihood of clinical trial success. AI provides highly accurate and fully reproducible analysis of those endpoints, resulting in a more precise patient snapshot and better ability to make eligibility decisions.

As AI-enabled diagnostics enter the clinic, the accuracy and precision that AI provides is ushering in a new era in precision medicine. And the latest activity in this space – including the integration of assay development with AI-driven endpoint assessment – is accelerating our journey towards this future.

Clinician well-being

Several trends are impacting work experience and well-being for physicians. Among them, an aging population with increasing patient case load, rising quantity and complexity of information at the individual patient level with the need for more complex reporting and cost pressures at the health system level resulting in additional stress and strain.

Meanwhile, in specialties such as pathology, the growing demand for specialists stands in stark contrast to a negative growth rate for trained specialists. These trends are manifesting in a growing physician burnout crisis that can only be addressed through transformational change, and technology will be part of the solution.

In pathology, digital pathology workflows are simplifying the workflow by eliminating inefficiencies associated with manual pathology (such as shipping of slides), speeding up the evaluation of cases (such as flagging for the presence of tumor), improving accuracy and repeatability of certain assessments (for example, counting the number of nuclei in glioblastoma) and enabling broader collaboration across institutions and borders through cloud infrastructure.

Ultimately, AI unlocks the pathologist’s time to operate at the top of their license by focusing on the most important and complex cases. The ability to utilize such expertise in daily practice will likely help attract more trainees to pathology while enabling their retention in the workforce because of improved and flexible workflows.

Lower costs

The relatively high investment cost for requisite infrastructure is a key barrier to widespread adoption of digital pathology and AI. While the return on investment may be meaningful at the level of individual laboratories, health system wide investment will have markedly greater impact.

Even adoption at the single large healthcare organization level will have near- to medium-term cost benefits. Digital pathology and AI will help laboratories improve productivity and remain competitive by addressing trends such as pathologist scarcity, cost related to inefficiencies and the ever-increasing complexity of diagnostic reporting.

Improved patient experience

In the era of precision medicine, patient experience hinges on the timeliness and accuracy of both diagnosis and treatment. It means safety from preventable medical errors. It also means having an understanding of disease prognosis and the likely impact of disease on one’s quality of life. As outlined above, these needs are at the core of how AI-powered pathology is transforming the diagnosis and treatment of disease.

In addition, generative AI-powered pathology chatbots and visual language models promise to assist pathologists in creating more standardized and patient-friendly pathology reports, ensuring that all critical therapy guiding information is included and adequately communicated.

Health equity

Achieving equity in health requires us to holistically address the drivers of health outcomes, of which the social determinants of health are the most important factor. It is not enough to have the best diagnostics and therapeutics if access to them is limited or indeed if the modifiable drivers of disease in society are not addressed. Steps have been taken in this direction in pathology, and this remains one of the key next frontiers for pathology and healthcare broadly.

In pathology, the creation of data sets that represent the diversity of the population, and leveraging these in drug and diagnostic development will address this key aim. Additionally, the distribution of digital pathology infrastructure will help to solve disparities in access to pathologist services and provide patients with access to the diagnostics and therapies that they need wherever they are. Other important interventions transcend pathology, and these include access to education, primary care, good nutrition, and clean living environment.

AI is already transforming translational research, clinical trials and the experience of both physicians and patients. There is every indication that we will continue to see growing adoption of this technology in research and in the clinic. A health equity-informed approach will help us ensure that patients everywhere can get the right treatment at the right time.

Eric Walk, MD, is Chief Medical Officer, Ben Glass is VP of AI Product & Translational Research, and Tafadzwa Muguwe, MD, is Sr. Director of Biopharma Operations at PathAI.

Return to AI BEYOND the Hype - March/April 2024 COVERstory.

More for you

Loading data for hdm_tax_topic #better-outcomes...