The convergence of human and artificial intelligence
Artificial intelligence will not replace human intelligence, especially in medicine. Diagnosis and treatment will remain a human endeavor. But AI will be an indispensable tool helping human intelligence effectively deliver better quality healthcare.
Medical Intelligence is a new discipline, converging human and artificial intelligence.
If It’s Just a Formula, How Hard Can It Be?
Healthcare is experiencing increasing developments and applications for artificial intelligence and machine learning. Not long ago, artificial intelligence was mostly theoretical because the processing power and the quantity of data required to give reliable real-world results simply did not exist. Now, a theoretical construct has become reality.
However, something’s wrong. A substantial quantity of data does not seem to more accurately determine diagnostic and treatment success. AI systems used to identify key variables and outputs for effective computer-based diagnosis and treatment recommendations are not much more accurate than humans. Why? Humans and medicine are not a simple algorithm that is solved easily.
Skynet Does Not Work
While there’s reason to be optimistic, useful real-world applications are still a struggle. The convergence of human and artificial intelligence is not straightforward. It is not simply large quantities of data manipulated with clever formulas. Data quality matters. Understanding and parsing useful data, merging inputs, and identifying similar characteristics build up useful data sets. From these data sets, powerful machine learning programs building useful neural nets and AI-based tools are finally giving us effective diagnostic tools and therapeutic outcomes. It’ a long journey and we are only at the outset.
AI is not Skynet. It does not make independent and uncontrollable decisions while ignorant humans stand by. Artificial intelligence is a tool enabling more effective human intelligence for application, judgment, and recommendations. AI will never substitute for human judgment, but it will augment and empower better human decision-making.
The data generated by individual human beings can be overwhelming, challenging the ability of health care professionals to understand it in any reasonable fashion. Each patient has multiple internal systems, performing multiple independent and interrelated functions — a genome, physiome, microbiome, proteome, epigenome, and many other complex layers and systems. Sensors and other devices can now receive continuous data from almost every one of these bodily systems. For example, we are now examining gut microbiome allowing us, among other things, to understand that we each have a unique set of private foreign species impacting our health and body — enabling the potential for truly effective individualized precision medicine.
All these different layers, defined for each person, generates an extraordinary amount of data. The hope is that gaining an understanding from this immense treasure trove will allow extraordinary advancements in preventing and treating illness, as well as managing new emerging illnesses, such as our current coronavirus pandemic.
· More unique diseases will emerge…
o AI strengthens diagnostic capabilities, effective treatment choices, and enforces our predictive and developmental arsenal.
So, Are More Zeros Better?
How much data? Start with a number then add 27 zeros… and counting. Yes. A lot. Perhaps too much.
For this overwhelming amount of data to be meaningful, we need the best-developed tools generating useful output. AI is the only viable solution because this technology has the promise of handling this flood of data while giving more accurate diagnostic tools and therapeutic outcomes.
· Currently, data is overwhelming healthcare and medicine, more often creating diagnostic errors, generating substantial waste via needless testing and procedures, and not enough focus on the patient’s true needs and actual condition.
AI is often misunderstood as something that only benefits a narrow group of specialists. The benefit is not only to medical specialists, such as radiologists but all medical practitioners from nurses to pharmacists to paramedics. Neural networks can be trained to provide extremely accurate diagnostic recommendations, whether that is via radiology, ophthalmology, or cardiology. Exceptional tools are becoming available and more highly refined. For example, in radiology, we now have the power of using perhaps millions of images to focus in and see things no human expert can, augmenting — not replacing — human judgment. This quality and accuracy will bring enormous benefits to many demanding areas; two examples include cancer and pre-diabetic diagnoses.
The overwhelming benefit is that it raises the bar for all practitioners.
· A minimum level of quality medical care can available globally.
· The higher standard for diagnostic accuracy, therapeutic recommendations, and overall care from this mass of data gathering will improve overall health and wellness everywhere.
· Applied effectively, these tools also drive down overall healthcare costs, diagnostic errors, and unnecessary procedures.
· Greater accuracy eliminates needless testing and procedures significantly and delivers effective care more quickly.
· Diagnosis is more immediate, recovery times faster, care more available, and overall expenses reduced.
Star Trek was Right Again
An interesting example of practical application, eerily close to Star Trek’s tricorder, as demonstrated by Eric Topol of the Scripps Institute, is a cardiovascular scanner attached to a smartphone that generates a sharp image of the heart and its functioning. This image can be taken by a smartphone of any part of the body, not just the heart (except the brain), anywhere in the world. Now, this information can be shared immediately with experts in any location and a smartphone ultrasound can be used to diagnose many conditions, including cardiovascular issues, pneumonia, or other life-threatening conditions anywhere, including in the developing world where no real medical infrastructure exists.
There are many similar examples representing information, diagnosis, and treatment recommendations that can be easily distributed globally.
· It is this game-changing paradigm — a smart device with intelligent software connected to a global network — encapsulating a phenomenal worldwide opportunity to fundamentally change medical science and its applications. We have an immediate and accurate tool that can be distributed globally.
Fundamentally, AI tools are available to any local facility, enabling a quick and efficient network for testing, sharing data, and generating faster and more accurate diagnoses. Its power can be applied broadly. One example is applying it to a rapidly increasing healthcare scourge, diagnosing diabetes. Using a retinal scan, which can be done locally on simple equipment (the magic is in the software), the data is immediately available and shared with medical professionals. Since no medical professional needs to be on-site, this is a service that can be applied broadly and immediately — and enable much better medical outcomes via early and effective diagnosis.
· Software platforms allow more effective and earlier diagnosis, leading to effective treatment and enhancing each patient’s quality of life.
Medical Intelligence is Medicine’s Future
These software, hardware, and network advancements combined with AI-created tools enable virtual global medical systems treating many more patients more effectively. Powerful neural networks developed through machine learning programs can give medical practitioners real-time tools providing quality healthcare sooner and much less expensively.
The challenge is there is so much data resting in different data sets that sharing and collaborative work is a major obstacle to developing useful and accurate AI tools. The convergence of data, software, artificial and human intelligence is essential and must not be hindered.
Medical Intelligence is a global game-changer. The need is critical, the developmental foundation exists, the technology is poised to perform, and essential data is more available. It represents one of the most powerful ways to improve global healthcare, wellness, longevity, and the quality of life.
Many thanks to Eric Topol (https://vivo.scripps.edu/display/TopolEric) and Andrew Ng (https://profiles.stanford.edu/131927) for their contributions