In the realm of science fiction, we often see robot doctors replacing human ones. The reasoning is that artificial intelligence (AI), which is based on cold, hard facts and calculations, can dramatically reduce the chance for human error -- now possibly the third leading cause of death in America after cancer and heart disease. According to a recent study in the Journal of Patient Safety, 210,000 to 440,000 patients die annually due to preventable harm in hospitals -- a startling jump from the 98,000 preventable deaths estimated by the Institute of Medicine in 1999.
Now that the HITECH (Health Information Technology for Economic and Clinical Health) Act is forcing hospitals to finally upgrade their aging tech, AI is gaining ground in health care settings. Therefore, let's take a look at some exciting new technologies from IBM (NYSE:IBM), Google (NASDAQ:GOOGL), and Nuance Communications (NASDAQ:NUAN), which could one day make AI doctors a reality.
Dr. Watson gets smaller, faster, and smarter
IBM's supercomputer Watson, best known for its appearance on Jeopardy!, is the most advanced prototype of a virtual doctor. Watson was programmed with the education of a third-year medical student between 2011 and 2013. In March 2012, Watson processed and analyzed Memorial Sloan-Kettering Cancer Center's extensive patient records, histories, and research -- a task it easily handled with its ability to process 200 million pages of text every three seconds.
Since then, institutes like New York Genome Center, Maine Center for Cancer Medicine, and New York's Westmed Medical Group have pitched in to enhance Watson's oncology database with additional information about certain types of cancer such as glioblastomas (brain tumors) and lung cancer.
When a doctor inputs a sentence describing the patient's symptoms on a connected iPad, Watson scans its accumulated records, cross references them with a patient's EHRs (electronic health records), and delivers treatment options based on those results. Additional input regarding symptoms will result in Watson narrowing and refining its treatment suggestions.
Watson was originally the size of a bedroom, but is now only the size of three pizza boxes, making it much easier to install in hospitals. With further development, Watson could perhaps get even faster and smaller in the future.
Google's DeepMind and Glass
Meanwhile, Google has already demonstrated a strong interest in the health care sector with its biotech subsidiary Calico and medical devices such as smart contacts. However, two other businesses -- Google Glass and DeepMind -- suggest that Google may play a crucial role in pushing AI technologies like Watson to evolve.
Imagine if Google Glass, which already has lots of potential uses in hospitals via EHR and augmented reality apps, could also synchronize with Watson. Doctors could naturally talk to Glass like a virtual assistant, and recommended treatments would show up on the display. Unlike the iPad, it would also be a completely hands-free experience.
In January, Google acquired DeepMind Technologies, a London-based AI company, for more than $500 million. On the surface, acquiring DeepMind was a great way for the company to improve its Google Now voice search algorithm for mobile devices. But the acquisition also implied that Google was trying to beat Watson at understanding processed data.
In a recent article from The Guardian, Google's Director of Engineering Ray Kurzweil claimed that whereas Watson uses "pattern matching" to process its data, Google intends to teach computers "to understand the meaning" of the processed documents. Therefore, Google could soon follow Watson's footsteps to create a rival medical student.
Nuance's CLU serves as a foundation for medical AI
Last but not least, it's easy to overlook Nuance, the voice recognition software company best known as the creator of Apple's (NASDAQ:AAPL) Siri. Unlike IBM and Google, Nuance already has a mainstream customer base with a proven medical AI system.
That system, known as Clinical Language Understanding (CLU), combines Natural Language Processing (NLP), speech recognition, and medical AI to create custom solutions for health care informatics. In a nutshell, CLU allows doctors to dictate a patient's information via a natural narrative, and Nuance's products -- such as Clintegrity 360 and Dragon Medical 360 -- input that data into EHRs.
While that system isn't as awe-inspiring as IBM or Google's AI programs, the ability to process a physician narrative into EHR data is just as important as an AI program's ability to produce medical recommendations based on accumulated clinical data and EHRs. By helping physicians input data more efficiently into EHRs, the adoption of EHRs will increase, producing a larger pool of patient data, which can be accessed by massive AI programs like Watson.
The Foolish takeaway
A recent study at Indiana University found that machine learning, used for voice recognition and credit card fraud detection, could cut health care costs and improve patient outcomes by nearly 50%. The machine learning model simulated various treatment paths, then continually eliminated or changed them when new information became available -- much like Watson's approach.
In closing, AI doctors probably won't replace human ones in the near future, but they can be used as assistants to remember and do what the human mind can't. That could result in a dramatic reduction of preventable deaths in hospitals, thanks to industry pioneers like IBM, Google, and Nuance.