Two Cents on AI in Healthcare

Thanks to the emerging of AI, huge progresses have been made in the past decades in healthcare industry. Not only we can see people surging into this area for the scientific advances, but also infinite business opportunities. Let’s take a look at the numbers: [1]

– According to venture capital firm Rock Health, 121 health AI and machine learning companies raised $2.7 billion in 206 deals between 2011 and 2017.

– A potential of $150 billion in annual savings for U.S. health care by 2026 with the advance of AI.

 

So how does AI indeed accelerate the healthcare domain?

Off-load repetitive human labors

A general application of AI is Chatbot, which specifically attributes to natural language processing, a subset domain of AI. While it’s widely used for lots of customer facing scenarios like sales and customer services, hospitals could also adopt it for patients’ self guides/evaluation. According to a report [1], traditionally, nurses and physicians spent more than half of their time (51%) on non-patient-care-activities. Now with the help of AI, nurses and physicians can be more freed up to work on their specialties.

Offer faster and more accurate analysis

It’s now generally accepted that machines outperform human beings in many areas, especially computation. Trained with big data, image processing of AI could now have a better precision than humans in analyzing X-Ray and CT scans, along with shorter analysis time. [2] Although experienced doctors could eventually come up with similar accurate rate of clinic image analysis, huge amount of doctor time could be saved from preliminary diagnosis of bone fraction, or even tumors. It’s a double win.

Scale-out personalized treatment plan

We’ve all experienced the below scenario: get prescriptions from doctor; stick to the instruction of at least 2 weeks; come back again for further evaluation; new prescriptions if applicable…And most of the time, doctors make educated-guesses on the dosage, depending on patients’ condition. Of course such procedure is not ideal. But that’s what we have being doing given limitation of doctor-resources. Now with AI, researchers have made a day-to-day evaluation followed with a personalized treatment plan possible. Such dynamic recovering process also comes with scalability. Not too long, everyone would be able to have his or her personalized health assistant. [3]

 

Future Perspective

Now we have everything on track: all above stated is on the way of boosting healthcare industry. Even if they haven’t made the radical transition right now, it wouldn’t take too long to complete. While the achieving of AGI (Artificial General Intelligence) is still under debated, and no transformative progress has been made in that area ever since the proposal of Turning test, one outcome is for sure achievable: a super machine with a collective of weak AI. Yes, AGI with emotions or super power might still only exist in sci-fi, and emotional support is also one of the needs in the healthcare. But in transition, if a control plane could operate on multiple modules, each designated for 1 area of diagnosis, then a self-operated super-power-doctor-substitution is nothing far-fetched.

 

[1] https://hbr.org/2018/05/10-promising-ai-applications-in-health-care

[2] https://medium.com/stanford-ai-for-healthcare/superman-isnt-the-only-one-with-x-ray-vision-deep-learning-for-ct-scans-290aaa7ba5c1

[3] http://stm.sciencemag.org/content/8/333/333ra49