AI’s potential to deliver better patient care

By Russ Altman

This is a lightly edited excerpt of testimony recently provided to the U.S. Senate’s Health, Education, Labor, and Pensions hearing "AI's Potential to Support Patients, Workers, Children, and Families."

The use of AI to analyze medical images — detecting cancer in lung X-rays or melanoma in pictures of skin — was the first to herald the revolution of AI in medicine. However, there are now myriad ways AI is being applied at various points of the patient care journey, from scheduling and triage to documentation support and patient communications. Properly validated and deployed, these tools will help care and treatment become more personalized, fair, and affordable. Let me highlight three areas where AI has the potential, and — in many cases — is already used, to transform patient care.  

First, like many, I am excited about AI’s ability to augment clinical diagnosis and treatment. Over 75% of FDA-cleared AI tools are for medical imaging — reflecting strong performance in pattern recognition. However, AI's use in healthcare can be much broader. At Stanford Health Care, as at many systems, we are seeing the careful rollout of tools that both read images and analyze patient records and clinical notes to flag likely conditions, such as cancer and peripheral artery disease, for timely follow-up. Used this way, AI helps prioritize patients, surface risks earlier and focus clinicians’ attention where it matters most, without replacing the professional judgment. My group has prototyped an AI system that uses electronic medical records and ancillary data to predict a patient’s risk for every disease; our work is a first step toward a dashboard where a physician can not only see the patient’s current diseases, but also those most likely to develop in the future. We have also used AI to process genetic information to choose the medication most likely to work for an individual.

Second, healthcare AI applications can improve the relationship between physicians and their patients. Hospitals and physicians across the country are rapidly adopting AI scribe tools that, within minutes, can generate transcripts and summaries of clinical conversations. Initial pilots have not only shown that these tools significantly reduce the burden of after-hours documentation work, which often contributes to physician burnout, but also improve the physician-patient relationship. Many patients are already noticing that their physician is paying more attention to them during clinic visits, while physicians feel they can provide more effective and personalized patient care. The days when physicians pay more attention to a computer screen than to their patients may soon be over.

Third, AI can increase patients’ understanding and control of their care using AI tools. Patients are eager to better understand their diagnoses, the progression of diseases, treatment plans, and medication side effects. We know that patients are going to general chatbots for medical information, much as they have used search engines in the past. Meanwhile, we physicians are often not as good as we think at explaining complex conditions in simple terms. Specialized medical LLMs that explain lab results in plain language or provide patients with an estimate of the likelihood that they will develop a certain disease could be powerful tools for transforming patients’ ability to feel empowered in their healthcare.

While I am optimistic about these applications, we will only fully realize their benefits if healthcare systems build interdisciplinary teams that thoroughly evaluate tools for clinical effectiveness and safety.  

Read the full testimony here.

Russ Altman is a professor of Bioengineering, Genetics, Medicine, and Biomedical Data Science. He is also a senior fellow at the Stanford Institute for Human-Centered AI. 

_{*The opinions expressed in this column are those of the author and do not necessarily reflect the views of HealthPlatform.News.}