The impact of artificial intelligence, or more specifically machine learning, is being felt in every industry sector, but perhaps nowhere more so than in healthcare, where AI funding hit historic highs in 2018, according to CB Insights.
The term AI is commonly used by the media and others to describe a computer-generated solution that is as good, or better than, a solution that could have been produced by a human. That often includes digital health tools that use algorithms programmed by researchers and clinicians. Machine learning is a subset of AI that uses neural networks to simulate or even expand on the level of data analysis that human minds are able to achieve. Deep learning is where software learns to recognize patterns. And these tools are already transforming diagnostic imaging.
CB Insights reports that since 2013, $4.3 billion in private equity has been invested in healthcare AI startups across 576 deals. That’s more than AI startups in any other industry have taken in.
In a way, healthcare and AI are almost made for each other. The healthcare sector produces tons of data, but most of it is not being leveraged to provide the kind of insights it potentially could. The hope is that AI will be able to sort through this mountain of information to provide novel insights to improve the treatment or enable the prevention of disease.
In this post, we have rounded up some of the most promising applications for AI/ML in healthcare and examples of companies that are making it happen.
Diagnostic imaging is one place where AI has made significant inroads. According to Frost & Sullivan, there are 114 startups working on AI for medical imaging. CB Insights identifies diagnostics as a key driver of healthcare AI deals accounting for more than $200 million of investment through the first half of 2018. There are now AI-assisted tools to diagnose or aid in the diagnosis of conditions ranging from diabetic retinopathy to cancerous lesions to patients at risk of strokes.
That’s not surprising. Medical imaging generates large and similar datasets that, unlike data collected by wearables, is stored in a clinical setting and therefore easily accessible to researchers. AI’s ability to recognize patterns has been harnessed to help clinicians identify risks and reduce the incidence of false positives that may require additional testing or unnecessary procedures.
Radiologists have now incorporated this technology into their workflow. Equipment manufacturers are already working to integrate AI technology into the next generation of imaging devices.
Cardiogram is an example of the next generation of AI-based diagnostics. The company’s goal is to take diagnostics out of the clinic and put it into the hands of anyone with a wearable device. Its tool currently allows users to track sleep and fitness and may eventually be able to predict a user’s risk of a serious cardiac event.
Drug Research and Development
Drug research is a notoriously inefficient process. The hope is that AI and ML can be leveraged to change the paradigm in this space. Interest in drug R&D, as measured by investment, has leapt in the past two years. In the first half of 2018, CB Insights recorded 20 AI drug discovery deals—equaling the total number of deals from 2017.
AI-based tools can predict interactions and identify possible side effects of a specific drug candidate, significantly reducing the time it takes to develop a drug or eliminating drug candidates early in the process. These tools can also be used to identify genetic variations that might affect a drug’s efficacy in certain populations—a key to personalized therapeutics.
Big pharma companies are also making significant investments in AI drug discovery deals either through strategic investments, partnerships or outright acquisitions. Most are using AI platforms in the hope to accelerate the search for successful drug candidates.
One example of this trend is Evidation, a health and measurement company that uses an AI-driven platform to help life sciences and healthcare companies understand how everyday behaviors and health interact. The company analyzes consented real-world data from smartphones and connected sensors, such as those in wearables and medical devices, to help with the identification, treatment and monitoring of disease.
At this point, most AI therapeutics are actually driven by algorithms. Most are designed to manage chronic diseases, particularly diabetes. Omada Health received full recognition of its diabetes prevention program from the Centers for Disease Control and Prevention in May 2018.
Mental health is also proving to be another critical use case for AI.
With the high cost of mental health therapy, a shortage of clinicians and significant populations that lack access to mental health services, AI-based mental health chatbots focused on cognitive behavioral therapy (CBT) are looking to fill that gap.
One example is Woebot, which is developing an AI-powered companion that leverages CBT and coaching techniques. An early study showed that patients had better outcomes when CBT was delivered through a “conversational agent” in the form of a bot than when the same material was delivered via an e-book.
For a roundup of top healthcare AI developments, visit CB Insight’s “The AI Industry Series: Top Healthcare AI Trends To Watch.”
Originally published February 6, 2019 on Fenwick's Life Sciences Legal Insights blog.