AI in Healthcare: Use Cases for 2020
AI is high on the hype cycle, not always delivering exactly what’s promised and sending businesses scrambling. Many AI use cases aren’t artificial intelligence at all, but instead, fancy processing used to handle big data. One area where AI is transforming human lives the way hype suggests may be invisible if you aren’t involved in a specific population segment — healthcare. Many of the most compelling use case examples of AI in healthcare are happening right under our noses in medicine, lifestyle management, and diagnosis, operating in a way that’s both disruptive and barely noticeable.
[Related Article: 15 Open Datasets for Healthcare]
That seems counterintuitive, but the most disruptive trends in specific fields are nearly invisible to recipients of the advance. You may not know your doctor is using better data to diagnose, treat, and monitor you, but it’s happening. You may not realize that your new prescription is the result of accelerated drug discovery through AI, but it’s here. Let’s take a look at some of the most exciting examples of AI-driven healthcare for 2020.
End to End Adoption
One of the most significant differences with healthcare is that adoption is happening on both sides of the table. Not only are developers in the healthcare field eagerly taking advantage of what AI can do, but their target customers, i.e., doctors and patients, are also willing adopters. While healthcare adoption is slower than other industries, it’s silos and high privacy walls causing it, rather than skepticism.
AI can reduce healthcare costs by taking over labor heavy human tasks and reducing or eliminating areas prone to human error. It’s also fostering a better “bedside manner,” allowing physicians to continue to see more patients while remaining fully informed about what’s happening in their patients’ lives and medical progress.
AI in healthcare is a testament to the concept of augmented intelligence, rather than purely artificial. The expertise of doctors, researchers, hospital staff, and others provides the technical bases for AI to learn. In return, AI lifts the heavy load of big data processing and pattern recognition, allowing healthcare professionals to return to what they do best, problem solve, and innovate.
AI Use Cases for 2020
Let’s take a look at this relationship in action.
Reducing Error in Diagnosis
Medical errors could be the third leading cause of death in the United States. While it’s hard to get an official handle on that number, long hours, too many patients, and a lack of access to data can be a disastrous combination when it comes to diagnosis.
Enter AI. Humans aren’t good at big data, and we aren’t always great at understanding subtle patterns. Big data isn’t something our brains naturally comprehend, but computers are excellent at these tasks.
For example, PathAI is using machine learning to technology to help pathologists catch more cases of cancer earlier. Enlitic uses AI to analyze unstructured data from radiology reports, blood tests, EKGs, and other sources of information to help advise doctors in real-time.
We’re also using computer vision in the form of AI-enhanced microscopes to scan for deadly blood pathogens in samples at a far faster rate than human staff could ever dream, allowing diagnosis with a 95% accuracy rating despite the increase in speed.
As companies continue to build more efficient programs, doctors and medical staff can increase the speed of diagnosis with little effect on day to day operations. Errors often come with pressure for speed, but here, we get the best of both worlds.
Drug Discovery
Drug research is costly and slow. It costs around $2.6 billion to put drugs just through clinical trials with a 10% success rate. Those are terrible odds, but until the advent of AI, we didn’t have much choice.
In 2007, an AI program named Adam was able to scour billions of data points about yeast, producing 19 hypotheses concerning gene codes within that yeast. Nine were completely new, and only one was wrong. Adam’s counterpart Eve discovered that triclosan could potentially combat drug-resistant malaria parasites through the same data analysis.
AI enables researchers to reduce human labor working on hypotheses and apply processing power to multiple hypotheses at once. AI can even be used to discover new materials and use patient material to drive predictive hypotheses.
Drug discovery has always been trial and error, a land of happy accidents and decades-long research. Those stories about discovering things like Rogaine or Viagra? Great stories, but Rogaine researchers were searching for hypertension drugs, and Viagra was in testing to treat heart-related chest pain.
Even penicillin, discovered entirely by accident, took over a decade to come into being after that first encounter with a moldy petri dish. Basic biology dictates that we have to be faster than that to keep up with drug discovery and evolution itself.
AI could provide support for speeding up the process while making it cost-efficient. We don’t have to use trial and error and can analyze data we already have to see things that have been right in front of our faces the whole time.
Work Flow and Documentation Assistance
Natural language processing is unlocking unstructured data directly from doctor’s notes, and it could help reduce errors and speed up the transfer of information. Documentation is a vital part of patient history, not just for the patient but for the entire community.
Without proper documentation, historical data can be lost or obscured. Missing just a single detail can be a deadly occurrence. When relying on a history of note-taking, extracting vital information can be a slow process. It’s estimated that data loss can cost the industry around $100 billion per year. With AI, we might change that.
Programs that sift through this mountain of data can inform doctors of patient risk, operationalize hospital workflows, and even predict the best ambulance routes. The more AI takes on the burden of complex administration tasks, the better staff can handle the uniquely human side of medicine.
Lifestyle Management
Follow up care is critical for maintaining health and wellbeing across a range of healthcare outcomes. The problem is that it’s expensive and out of reach for a large swath of the population. Now, companies are integrating AI and IoT to close that gap.
Wearables are probably one of the biggest promises for lifestyle management. This goes beyond your smartwatch measuring your heart rate. Instead, these wearables can detect small changes in heart rates, alert a doctor to the change, diagnose potential issues, and advise the patient to seek medical help, all from a single loop.
Other advances are friendly chatbots, helping people like seniors remember to take medications, and interacting to help detect early signs of depression, for example. These advances bring the doctor’s presence to the patient.
Even more intriguing is the potential to improve communication between doctors and patients directly. Although barriers exist due to healthcare’s wall of privacy, AI could analyze real-time conversations, for instance. AI is a better predictor of human behavior than humans are and could detect misunderstandings, offer treatment suggestions, and analyze the tone of conversations.
Read more data science articles on OpenDataScience.com, including tutorials and guides from beginner to advanced levels! Subscribe to our weekly newsletter here and receive the latest news every Thursday.
