• Hypertrophic Cardiomyopathy (HCM)
  • Vaccines: 2023 Year in Review
  • Eyecare
  • Urothelial Carcinoma
  • Women's Health
  • Hemophilia
  • Heart Failure
  • Vaccines
  • Neonatal Care
  • Type II Inflammation
  • Substance Use Disorder
  • Gene Therapy
  • Lung Cancer
  • Spinal Muscular Atrophy
  • HIV
  • Post-Acute Care
  • Liver Disease
  • Biologics
  • Asthma
  • Atrial Fibrillation
  • RSV
  • COVID-19
  • Cardiovascular Diseases
  • Prescription Digital Therapeutics
  • Reproductive Health
  • The Improving Patient Access Podcast
  • Blood Cancer
  • Ulcerative Colitis
  • Respiratory Conditions
  • Multiple Sclerosis
  • Digital Health
  • Population Health
  • Sleep Disorders
  • Biosimilars
  • Plaque Psoriasis
  • Leukemia and Lymphoma
  • Oncology
  • Pediatrics
  • Urology
  • Obstetrics-Gynecology & Women's Health
  • Opioids
  • Solid Tumors
  • Autoimmune Diseases
  • Dermatology
  • Diabetes
  • Mental Health

Medical Records Can Be Messy. One Company Says AI Can Bring Clarity

MHE PublicationMHE February 2024
Volume 34
Issue 2

xCures says artificial intelligence and machine learning can help make sense of disparate medical records.

© ipopba - stock.adobe.com

© ipopba - stock.adobe.com

Blame it on Deep Blue.

Ever since IBM’s Deep Blue computer defeated chess champion Garry Kasparov in 1997 humans have wondered — and fretted about — whether computers could outperform humans in a wide range of areas.

At xCures, a California-based healthcare technology startup, the question took the form of a competition between artificial intelligence (AI) and expert panels of physicians (known as tumor boards) to see which could make better treatment decisions for patients with advanced cancer.

xCures built an AI-based decision support application, and soon it was able to match the performance of human tumor boards, Mika Newton, CEO of xCures, said.

“I get asked, ‘Would you still want a tumor board with humans on it?’ ” Newton told Managed Healthcare Executive. “And the answer is, yeah, maybe there’s some magic in there sometimes, though I think that actually happens less than people expect.”

The humans versus computers question occupied much of xCures’ first two years after its founding in 2018. However, Newton said it also exposed another problem that has led to the company’s growth since then. “And that problem was access to sufficiently sophisticated medical records — medical data — to make those decisions,” he said.

Nonprofit origins

xCures has firsthand knowledge of the difficulty of making meaning out of disparate medical records, because it used to do it the old-fashioned way. The company grew out of Cancer Commons, a nonprofit organization that provides navigation and advocacy services for patients with advanced cancer. Such work has historically been human labor-intensive, relying on individuals to stay up to date on the latest trials and therapies and the details of individual patients’ cases. The work helped connect patients to cures, but it had logistical limits.

“They were starting to look at what was the type of technology that you would need to build in order to really be able to scale that nationally,”
Newton said.

To achieve scale, Cancer Commons launched a for-profit company, xCures. The challenge xCures faced was finding a tool that could give them all-comers data on patients with advanced cancer nationwide as well as the comprehensive medical histories of individual patients. “And we couldn’t find such a platform,” he said, “so we ended up building one.”

What xCures built is software that uses AI and machine learning (ML) to pull together all available medical records on a patient within 15 minutes, giving physicians the information needed to best advise patients.

“Then the last part is that the data [are] now sufficiently organized and structured so that you could do something useful with [them],” Newton said.

The average patient has 1,400 files from 30 different provider locations, he said. xCures provides reports that summarize the data and also link to the original files in case a provider wants to double-check a piece of information or dig deeper into it. In some cases, records are based on facsimiles or scans or include jargon specific to a particular health system or region, Newton noted. AI and ML can sort through those anomalies and make sense of such data.

Other companies

xCures is not alone in seeing the potential of AI to make sense of disparate medical records. Back in 2021, Sidhartha R. Sinha, M.D., of Stanford University in California, and colleagues developed an AI model to organize and display referral records for new patients. They then recruited a dozen physicians and gave them one set of records that had been organized by the AI system and one set of records in the standard, nonoptimized format. After reviewing the records, physicians were given a list of 22 questions to answer based on the records. The investigators found the AI-based system cut the amount of time it took to answer the questions by 18%.

“The AI system helped physicians extract relevant patient information in less time while maintaining high accuracy,” Sinha and colleagues said in a write-up of the results published in JAMA Network Open in July 2021. “This is particularly relevant in an era in which practitioners are confronting increasing volumes of EHR [electronic health record] data and the loss of face-to-face interaction with patients.”

In fact, results from a 2018 survey of physicians conducted by The Harris Poll for Stanford Medicine suggested physicians spend approximately 62% of their time allotted to each patient referring to EHRs.

Even as xCures has built out its technology platform, public perceptions of AI have changed rapidly. In 2022, the technology firm OpenAI released a public version of its ChatGPT chatbot, ushering in a world in which the public could use AI to create everything from travel itineraries to haikus and 10th-grade history essays. “I think that really sparked everyone’s imagination that AI could now move faster and we actually have enough computing power to do these really startling things,” Newton said.

It also raised concerns about potential harms that could come from AI. The straightforward nature of xCures’ service puts it in the relatively noncontroversial corner of the AI world; everything the company produces can be easily verified and its provenance documented.

Still, even relatively straightforward uses of AI in healthcare raise significant concerns. Saad Abdullah, Ph.D., of Mälardalen University in Sweden, and colleagues noted in a 2023 paper in Biomedical Materials & Devices that medical records are “seldom organized neatly” and are “often erroneous.” Abdullah and colleagues noted that healthcare datasets such as those used and created by AI systems raise significant privacy concerns and are also vulnerable to hackers and ransomware attacks.

They pointed out that a number of countries have enacted laws and regulations designed to protect patient privacy, but they said such laws can have unintended consequences.

“Because various laws passed by various countries make problems of collaboration and cooperative research more difficult, data privacy regulations established to solve this issue may restrict the quantity of data accessible to train AI systems on a national and global scale,” they wrote.

Newton said the healthcare sector is not unique in needing to navigate the intersection of AI and privacy. He noted that any time a person applies for a credit card, for instance, a host of personal financial records are used to verify the applicant’s identity and assess their creditworthiness.

When meeting with potential clients, Newton said their concerns are generally less about the technology and more about liability when errors occur. Specifically, if a physician misses a key piece of a patient’s medical history or makes a decision based on erroneous data, is the physician liable for any resulting harm or is the AI company liable? “That stuff is really hard to tease apart,” Newton said.

‘Scary idea’

Sometimes such mistakes are due to negligence or a lack of sufficient governance. Other times, mistakes are just mistakes. So far, Newton said he is unaware of any court cases testing such questions. However, he said similar issues arise regarding self-driving cars and that litigation over accidents caused by self-driving vehicles might create some legal clarity for the healthcare industry.

In the meantime, Newton said xCures plans to expand its market beyond cancer and begin looking for other types of illnesses in which its technology can make a meaningful difference. He said he understands the concerns some have about AI and healthcare.

“It’s just a scary idea,” he said, “because you don’t know what’s really possible or not.” Newton pointed to a recent article in a national newspaper questioning whether AI would be able to match the quality of human-made art, thus rendering it redundant. While it’s fun to ponder such questions, Newton thinks such juxtapositions might be missing the point.

“As a technologist, I’m just not sure it needs to go that far,” he said. “I think we should say, ‘These are the tools we have. These are the problems we want to solve. What are the right tools for the problem?’ ”

Jared Kaltwasser is a healthcare writer in Iowa and a regular contributor to Managed Healthcare Executive.

Related Videos
Video 9 - "Denial of Coverage in Fertility Care"
Video 8 - "Risks of Miscarriage and Multiple Births Associated with Fertility Care"
Video 7 - "Fertility Preservation: Egg Freezing Versus Embryo Freezing"
Video 6 - "Family Building Costs, Barriers, and Dropout Rates Associated with Fertility Care"
Video 5 - "Closing Payer Gaps and Improving Fertility Care Access"
Video 4 - "Increasing Employer Coverage and Maximizing Fertility Benefits "
Video 5 - "Relevance of NUTURE Study Findings for Patients, Payers, Providers"
Video 3 - "Improving IVF Success Rates & Utilizing AI in Fertility Health Care"
Video 2 - "Holistic Fertility Management and Payer Collaboration"
Video 5 - "Relevance of NUTURE Study Findings for Patients, Payers, Providers"
Related Content
© 2024 MJH Life Sciences

All rights reserved.