VentureBeat presents: AI Unleashed - An exclusive executive event for enterprise data leaders. Network and learn with industry peers. Learn More
Table of contents
Artificial intelligence (AI) is being applied across the healthcare spectrum — from administration to patient interaction and medical research, diagnosis and treatment.
What is healthcare AI?
Healthcare AI is the application of artificial intelligence to medical services and the administration or delivery of medical services. Machine learning (ML), large and often unstructured datasets, advanced sensors, natural language processing (NLP) and robotics are all being used in a growing number of healthcare sectors.
Along with great promise, the technology offers significant potential concerns — including the abuse that can come from the centralization and digitalization of patient data as well as possible linkages with nanomedicine or universal biometric IDs. Equity and bias have both also been concerns in some early AI applications, but the technology may also be able to improve healthcare equity.
Although deployment of AI in the healthcare sector has truly just begun, it is becoming more commonly used. Gartner pegged 2021 global healthcare IT spending at $140 billion, with enterprises listing AI and robotic process automation (RPA) as their lead spending priorities.
An exclusive invite-only evening of insights and networking, designed for senior enterprise executives overseeing data stacks and strategies.
Healthcare costs approached a fifth (19.7%) of the total U.S. economy in 2020 (an estimated 19.7% or $4.1 trillion). Over half of that spending, for the first time, was racked up by the government, where fraud is especially high.
Thus, the potential value of healthcare AI, from administration to medical AI is vast.
10 top applications of artificial intelligence in healthcare in 2022
Here are 10 of the top areas where healthcare AI use cases are being developed and deployed today.
1. Healthcare administration
Administrative expenses are estimated to comprise 15% to 25% of total healthcare costs. Tools to improve and streamline administration are valuable for insurers, payers and providers alike.
Identifying and cutting down fraud, however, may provide the most immediate return as ealthcare fraud can happen on many levels and be committed by various parties. In some of the worst cases, fraud may cause insurers to get billed for services not rendered or result in surgeons performing unnecessary operations to get higher insurance payments. Insurers may also get billed for defective devices or test kits.
AI can be a useful tool in stopping fraud before it happens. Just as banks commonly use algorithms to detect unusual transactions, and health insurers can do the same..
- McKinsey has identified the potential savings with algorithm-driven “smart audits” of incoming insurance claims.
- The U.S. government’s Centers for Medicare & Medicaid Services leads a Healthcare Fraud and Prevention Partnership to identify patterns across pooled databases.
2. Public health
AI is already being applied across the public health sector. Including
- ML algorithms are being applied to large public health datasets, and the CDC has compiled some of the many ways AI has been applied in analyzing public health for COVID-19 and beyond.
- NLP is being applied in public health contexts.
- Increasingly, diagnostic imaging data is being harnessed for population-level analysis and predictions.
- Lirio applies consumer data science and behavioral “nudging” techniques to creating “precision”, or personalized, nudges to prompt healthcare visits, medical compliance and the like.
3. Medical research
The applications for AI in medical research are also expansive. Examples range from new and repurposed drug discovery to clinical trials, including:
- Finding new drugs to treat conditions can be incredibly complicated . In silicon computer-aided drug design (CADD) is its own complex field.
- In some cases, the goal is to repurpose existing drugs. One recent example came when AI analyzed cell images to see which drugs were most effective for patients with neurodegenerative diseases. Neurons change shape when positively responding to these treatments. However, conventional computers are too slow to spot these differences.
- Pharma provider Bayer believes AI could enhance clinical trials by creating a virtual control group using medical database information. They’re exploring other AI clinical trial applications, too, that could make these investigations safer and more effective.
4. Medical training
AI may also alter how medical school students receive parts of their education. Including in cases like the following:
- One example gave students feedback from an AI tutor as they learned to remove brain tumors. The system had a machine learning algorithm that taught students safe, effective techniques, then critiqued their performance. People learned skills 2.6 times faster and performed 36% better than those not taught with AI.
- Organizations in the U.S. and the U.K. have also deployed AI-based virtual patients to facilitate virtual and remote training. That approach was particularly useful when the COVID-19 pandemic halted group gatherings. The AI supported practicing several skills, likecomforting distressed patients or delivering bad news.
5. Medical professional support
AI is also deployed to support medical professionals in clinical settings, including the following:
- AI is applied to support intake professionals in medical facilities. One Stanford University pilot project uses algorithms to determine if patients are high-risk enough to need ICU care or to experience code-related events or those that require rapid response teams. They assess the likelihood of those events occurring within a six to 18-hour window, helping physicians make more confident decisions.
- AI-based applications are being developed to support nurses, with decision support, sensors to notify them of patient needs and robotic assistance in challenging or dangerous situations among the areas addressed.
6. Patient engagement
AI is also deployed to support patients directly:
- Hospitals use AI chatbots to check in with patients and help them get necessary information faster. When Northwell Health implemented patient chats, there was a 94% engagement rate among those utilizing oncology services. Clinicians who tried the tool agreed it extended the care they delivered. Chatbots are able to check on patients’ symptoms, recoveries and more. Many people are also used to chatting by text, which increases adoption. Chatbots also reduce challenges patients may encounter while seeking care. People can use them to find hospitals or clinics, book appointments and describe needs.
- Estimates suggest that as many as half of all patients don’t take medications as prescribed. However, AI can increase the chances of patients taking their medications as they should. Some platforms use smart algorithms to suggest when health professionals should engage with patients about compliance and through which channels. Medication reminder chatbots exist, too. In a recent example, researchers collaborated and used AI to assist with finding the best medications for people with Type 2 diabetes. The algorithms helped choose the right options for more than 83% of patients, even in cases where the people needed more than one medication simultaneously.
7. Remote medicine
Telemedicine in the form of virtual doctor visits have become increasingly common since the COVID-19 lockdowns. In addition to those, AI is supporting other forms of remote medicine as well, including:
- VirtuSense applies predictive AI to remotely monitor and alert providers about high-risk changes that may precipitate a fall.
- Some facilities currently using AI for monitoring rely on it for conditions ranging from heart disease to diabetes. Hospitals also used this technology to oversee COVID-19 patients, making it easier to decide which could receive home care and which needed hospital treatment.
AI is also utilized for healthcare center diagnostics, including by:
- One AI system used to spot breast cancer can detect current issues and a patient’s likelihood of developing the disease in the next several years.
- Some applications of AI in healthcare detect mental ailments, too. Researchers have used trained algorithms to identify depressed people by listening to their voices or scanning their social media feeds, for example.
AI does not eliminate surgical issues, but it can potentially reduce them while enhancing outcomes for patients and surgeons alike. This is illustrated in the following examples
- A startup called Theator recently raised $39.5 million in a series A funding round. The company has an AI video solution built to help surgeons see what went wrong and right during procedures. They can then study the footage to make improvements for the future.
- Artificial intelligence applications in healthcare include surgical robots that are increasingly common in operating rooms. Many are minimally invasive and often achieve outcomes superior to non-robotic interventions. These uses of AI won’t replace humans’ surgical expertise. Though, they can work as surgeons’ partners, improving the likelihood of procedures succeeding.
10. Hospital care
Along with the above-described diagnostic use cases, clinicians also must meet patients physical needs and, more prosaically, stock supplies and deliver goods. AI-powered collaborative robots are starting to ease the burden. Gartner expects 50% of U.S. providers to invest in robotics process automation (RPA) by 2023. Some examples of RPA in hospitals include:
- One hospital recently deployed five robots named Moxie. These machines will proactively determine when nurses need supplies or assistance with lab test logistics. They’ll then respond before the provider’s workload gets too intensive.
Atheon provides robots that support not only medical functions, but tasks such as linen distribution and waste removal.
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.