AI in Healthcare: Advancing Medication Adherence, Virtual Care, Smart Monitoring, and Drug Discovery

Artificial Intelligence is being used to enhance the efficiency and automation of healthcare services. While the development of AI technology was once heavily questioned, we now find that big data technology is driving its progress, including in the healthcare sector. Let’s take a look at several AI applications currently underway with VCBeat.

Analyze Patient Behavior to Develop Personalized Oncology Treatment Plans

For example, two breast cancer patients may receive the same treatment regimen, yet their physical conditions may be entirely different.

One might be a marathon runner, while the other is a quiet book lover; one could be a smoker, and the other perhaps a health-conscious individual; one may be in their sixties, while the other is just turning forty. Such scenarios are common in our daily lives.

Therefore, considering the various differences, these two patients require two different treatment plans.

For scientists and physicians, the challenge lies in accessing specific patient information. Critical data are often buried within vast amounts of information, leaving clinicians with insufficient time—potentially up to a year—to sift through the deluge and extract the insights they need.

Consequently, many researchers are striving to leverage artificial intelligence approaches to overcome this challenge.

For instance, scientists at Carnegie Mellon University and the University of Pittsburgh are leveraging artificial intelligence to extract valuable information from electronic health records, diagnostic imaging, prescriptions, genomic data, insurance records, and even data from wearable devices, thereby developing healthcare solutions tailored to specific diseases and patient populations.

Researchers leverage big data to develop tailored medical protocols, control infectious diseases, and seek treatments for life-threatening conditions.

“The biggest problem we face now is that the system is not intelligent,” said Eric Xing, a professor in the Machine Learning Department at Carnegie Mellon University. “The data stored in the system is essentially static, whereas machine learning and artificial intelligence can extract useful information from massive datasets. You can think of it as an artificial brain taking over the work of a ‘dead’ storage system.”

He stated that Carnegie Mellon University and the University of Pittsburgh are collaborating with the University of Pittsburgh Medical Center on a project called the “Pittsburgh Health Data Alliance.” Over the next six years, the medical center will provide annual funding of $10 million to $20 million to researchers for this initiative.

Scientists are leveraging de-identified health data obtained from medical centers to explore more rapid and effective big data analytics, aiming to develop healthcare-related technologies and services that enable improved diagnosis, treatment, and communication tailored to individual patients.

“Every patient is a unique individual,” Xing added. “Take the simplistic view that breast cancer should be treated with either Drug A or Drug B. However, due to the uniqueness of each person’s lifestyle, living environment, and other relevant health factors, every individual is distinct. Artificial intelligence does not merely extract information from a single physician; instead, it aggregates insights from a large number of experienced clinicians, thereby enabling it to identify common patterns across diverse patient populations.”

Furthermore, AI software operates with significantly higher efficiency than the human brain, enabling faster identification of data patterns and similarities to assist doctors and scientists in discovering the most critical information.

For example, a certain treatment may be highly effective for a 50-year-old diabetic patient with an active lifestyle. In such cases, physicians can apply the same treatment to other diabetic patients who share similar characteristics.

Xing stated that his team is currently developing an app designed to provide users with healthy lifestyle recommendations and help prevent certain diseases. The app is expected to launch within the next year.

Philip Lehman, Associate Dean of the School of Computer Science at Carnegie Mellon University, told the author that this app leverages artificial intelligence to advise users on when to seek medical attention, which type of specialist to consult, and how to maintain their health.

“For example, people now typically use their smartphones to search for ‘how do I get to a certain place,’” Lehman stated in an interview. “In fact, the same principle applies to healthcare: ‘What can I do to feel better or live longer?’”

Lehman and Xing hope to develop prototypes for a range of products, from apps to machine learning tools and services, aiming to launch more than a dozen new products over the next five to six years.

A prominent company in this field is Welltok, which has received investment from IBM. Leveraging IBM’s “Watson” supercomputer, Welltok aims to realize its vision of engaging users through personalized activities. Its app, Cafewell Concierge, utilizes Watson’s natural language processing capabilities to better understand user needs, balance motivational incentives with warnings, and thereby achieve desired outcomes to provide feedback to users.

Virtual Medical Assistants: Improving Medication Adherence

For example, AiCure leverages mobile technology and facial recognition to determine whether patients take their medications on schedule. It then collects patient data via a mobile app and employs automated algorithms to identify medications and monitor ingestion. Patient data is transmitted in real time to clinicians through a HIPAA-compliant network, enabling physicians to verify adherence to prescribed regimens. Additionally, this technology can be used to detect adverse events.

Another example is Alme Health Coach, an app developed by Next IT, which delves into the reasons why people fail to take their medications on schedule. Although Next IT is still a newcomer to the healthcare services industry, it has previously developed a “virtual assistant” app to help consumers address issues in banking, retail, property management, and other sectors.

Generally, some AI components repeat users’ statements to clarify their intent. In contrast, Alme Health Coach is specifically designed and configured for particular diseases, medications, and treatments. It can synchronize with users’ alarms to trigger prompts such as “How did you sleep?” and remind users to take their medications on schedule. This approach aims to collect actionable data that physicians can use to better engage with patients (provided that patients are willing to share their data).

Track status, auto-reporting supports intelligent care

Automated Insights, an AI technology company, has partnered with Great Call, a mobile app developer, to integrate its natural language generation platform, Wordsmith. Family members and friends can access information about the device wearer through Great Call devices connected to the app. Primarily designed for elderly care, the app sends alert notifications when the wearer requires assistance. Additionally, the app features patented GPS positioning technology to track the user’s location.

Currently, the company has been acquired by Vista Equity Partners and STATS (Sports Information Technology Company). Leveraging Wordsmith’s automated writing capabilities, textual reports will be automatically generated for caregivers, detailing information such as location, activity routes, battery status, and device usage.

Intelligent Drug Discovery and Development

Biotechnology companies are also integrating artificial intelligence with big data to identify new drug compounds, such as Cloud Pharmaceuticals and Berg.

Berg leverages its proprietary Interrogative Biology artificial intelligence platform to study healthy human tissues, investigate the body’s molecular and cellular self-defense mechanisms, and elucidate disease pathogenesis, thereby utilizing AI and big data to predict potential therapeutic compounds targeting endogenous human molecules.

This approach offers numerous advantages. It not only makes targeted therapy the current trend in medical treatment, but also leverages the body’s own molecules to treat complex and refractory diseases such as diabetes and cancer, reducing the time and financial costs by half compared to developing new drugs.

Of course, Berg is not the only company in this field. Cloud Pharmaceuticals is focusing on R&D in this area and has raised $20 million in funding.

Furthermore, Johnson & Johnson and Sanofi are also leveraging the “Watson” supercomputer system—a computing platform capable of rapidly identifying relevant patterns within massive datasets—to support drug research and development.

Johnson & Johnson uses "Watson" to rapidly analyze scientific papers detailing clinical trial results, accelerating comparative effectiveness research on different treatments to expand the drug's applications across broader fields—a task that would otherwise require three people ten months to complete using conventional methods.

“Watson” can now recognize the languages of chemistry, biology, law, and intellectual property, empowering scientists with a unique ability to “communicate” with data—an advantage unavailable to others—which will accelerate breakthroughs in scientific and medical research.

Sanofi, meanwhile, leverages Watson to identify new indications for existing drugs; Watson organizes and screens toxicology data to help researchers determine which drugs are most suitable for application in new therapeutic areas.

Compiled by John Wang