The Enigma Behind the Rise of Artificial Intelligence (AI) in Healthcare

Amid dissensions, economic pundits agree on one thing: healthcare technologies are becoming more complex throwing challenges for doctors to handle them productively. This is where Artificial Intelligence (AI) is making its presence felt in healthcare to assist physicians to comprehend the big data. AI brings in increased productivity, higher efficiency that many times translate to lower costs and help handle volumes of tests and scans.

One of the vital aspects that is helping the rise of AI is the exponential growth of computational power of silicon chips (microprocessors) and at the same time computational cost is coming down. If not for this imperative aspect, AI itself would have added more burdensome technical complexity along with the cost challenges.

Unequivocally AI is establishing itself in all specialties of healthcare. Below are a few unique illustrations to show how AI is affecting multiple specialties and is becoming inseparable part of our healthcare system.

AI in Colonoscopy

A typical colonoscopy takes about upwards of an hour for the procedure. Surprisingly, even for a well-versed physician, the success rate of detecting benign polyps is not 100 percent. It means that a doctor could possibly miss a cancerous polyp during a colonoscopy, resulting in expensive procedures later, not to mention the pain and suffering that the patient has to endure.

This is where AI is playing a significant role. Using its “Object Recognition” capability, AI can identify patterns and is capable of comparing the scanned image against the millions of data inputs from patients around the world.

AI systems can also look at more than three hundred different features of polyps instantly. Doctors at Showa University in Yokohama, Japan, were able to scan for polyps with accuracy range of 71% to 98%. The AI accuracy range depends on the quality of data that is used for AI training and also on the quality of data that is captured during the colonoscopy itself. AI-assisted colonoscopy technologies are becoming popular as they are more reliable compared to those done by humans alone. AI is just a computer algorithm and its accuracy keeps increasing as it learns more from the big data which is on the rise around the world. In an AI-assisted colonoscopy, if a doctor misses a polyp, AI would send an alert so the doctor could go back and take a second look in the same procedure. This dramatically increases the efficiency of colonoscopy and substantially reduces overall cost by preventing patient readmissions.

AI to Manage Kidney Stones

According to researchers from Massachusetts General Hospital in Boston, USA, AI can be used to accurately detect and characterize kidney stones. This research uses “Neural Network” an AI technique that combs through the CT scans to characterize the stone for easy management. AI has shown 90 percent accuracy both in detection and composition analysis. This substantially increases the efficiency of the procedure. AI can even predict the likelihood of having a kidney stone again in the future, further reducing the healthcare cost of readmission as patients can take preventive steps.

AI to Manage Skin Cancers

Scientists at Stanford University created an AI diagnosis algorithm for skin cancer using a Google-provided public application programming interface—a set of routines, protocols, and tools for building software and applications. They made a database of nearly 130,000 skin disease images and trained their algorithm to visually diagnose potential cancerous cells. The algorithm was tested against twenty-one board-certified dermatologists. In its diagnoses of skin lesions, which represented the most common and deadliest skin cancers, the algorithm matched the performance of dermatologists, with a 91 percent accuracy rate. The data quality is highly important both during AI training and during the diagnosis. Here again higher efficiency, less doctor time, and fewer wrong diagnoses would reduce the complexity and reduce healthcare costs considerably.

AI in Surgery

AI surely is a significant add-on in the field of robotic surgery. Intuitive’s da Vinci surgical system is wellestablished in this market. Though human surgeons are still needed, the surgical procedure itself is robot-intensive. Surgeons have moved away from the operating table to remote control stations from where they control the operating robots. The surgeons still uses their surgical knowledge, but the precision cutting and stitching is now done by the machine.

Robotic surgeries help hospitals reduce their costs by shortening hospital stays and having fewer readmissions. These robotic procedures offer lesser patient discomfort and quicker recovery time. As of now robot-surgery is used only in gynecological and urological operations, but this technology has the potential to eventually spread across all areas of surgery.

There are many companies competing in the field of robotic surgery, each focusing on different aspects such as accuracy of the surgical instruments, better touch, better visualization of the patient’s anatomy, etc. As all these aspects are highly data driven, AI can decipher it better than humans.

Google has teamed up with Johnson and Johnson to come up with a robotic surgical device that uses AI to learn more about the anatomy of patients. Currently during robotic surgery, a surgeon only gets the video feed and they have to identify what those elements are. However, in this Google venture, AI can identify human anatomical parts using “Machine Learning” technique. This may not help a well-versed surgeon, but greatly helps new doctors especially during their training. And as these young, robotic surgeons progress into the future, they will depend more heavily on AI assisted technologies they have been trained with.

AI to Detect Brain Bleeds

IBM Watson and Israel-based MedyMatch Technology are coming together to use AI to help doctors detect brain bleeds resulting from head trauma and stroke. It uses machine learning algorithms that have access to machine vision and patient data to highlight areas of potential presence of cerebral bleeds. According to the American Stroke Association, stroke is the fourth leading cause of death and at the same time one of the top preventable disabilities in the United States. Any kind of automation that enhances efficiency substantially reduces healthcare costs.

AI to Detect Tuberculosis

Researchers at Thomas Jefferson University Hospital in Philadelphia are training AI to detect tuberculosis (TB). They are using two models—AlexNet and GoogLeNet— to identify TB on chest X-rays. The findings are published in the Journal of Radiology. AI is learning from a host of medical images to interpret radiographs for the presence of TB. This will have a huge positive impact in underdeveloped nations where there is shortage of physicians. According to the World Health Organization, in 2020, about ten million people fell ill from TB causing 1.5 million deaths. These kinds of automated technologies will enhance the global health standards.

Operational Efficiency in Hospitals

In busy hospitals, doctors operating on wrong patients or wrong part of the body and nurses administering wrong medications are not uncommon. These mistakes not only escalate the cost of healthcare but also cause painful suffering for the patients and even death. Many hospitals have adopted RFID (Radio-frequency identification) tags in bracelets to identify patients and get all the patient information. An RFID tag is nothing but an updatable bar code that would contain vital patient data. An RFID ‘reader’ is required to read these RFID tags.

Another technology Internet-Of-Things (IoT) is becoming very popular in healthcare today. It gives a digital voice to the electronic healthcare gadgets. IoT is nothing but a sensor/reader that is linked to the Internet. A digital thermometer that is linked to the internet is an IoT. A digital pacemaker that is connected to cloud is an IoT. An RFID reader that is connected to the cloud is an IoT. IoTs are becoming fairly common in healthcare to capture data and store it on the cloud. These cloud connected devices increase the operational efficiency in hospitals.

Consider a nurse giving medication to a patient. At the patient’s bedside, they can use a mobile computer and an RFID ‘reader’ to scan the patient’s RFID bracelet to get the patient’s name and date of birth and to bring up their medical record. The system then will provide a guided workflow with any instructions for administering drugs, underpinning the five “rights” of medicine administration— the right patient, the right drug, the right dose, the right route, and the right time. The nurse can also scan the barcode on the drug packet to confirm that the patient has no allergy to it. After the medication is administered, the system would then signal that the medication has been taken, with a note auto-created on the patient’s electronic medical records (EMR).

Though this tech-heavy approach ensures that the medication error is reduced to a minimum, it has dependency on the RFID tags, scanners, and the data they pull. It is possible that the data in the tag itself may be wrong because of a bad data entry! Or the tag itself is on a wrong patient! This results in the system trying to match a wrong patient with a wrong medication. This is where AI would come in to do the ‘Due Diligence’ evaluating the potential mismatches and false positives. AI could use ‘Facial Recognition’ technology to identify the patient and find any obvious errors in the system. If AI is carefully calibrated, it can help to ensure that the system is meeting the organization’s expectations. These kinds of vital validations of data would encourage organizations to invest in AI and relevant IoT gadgets.

AI in Tracking Shelf Life

In hospitals, it is essential to maintain the correct expiration date of all medicines and critical medical devices. The RFIDs/IoTs make the medical inventory monitoring simpler. In hospitals where there is a huge number of medical inventories, keeping track of their shelf life has become a non-issue, thanks to widespread IoT usage and AI algorithms that manage the inventories.

AI in Patient Monitoring

Tracking patient’s movement inside a healthcare facility and notifying where and when the patient is required to be can save a lot of time and substantially increases efficiency. For example, for angioplasty the goal might be to have the patient operated on within sixty minutes. By giving the patient a wristband with an IoT tag that confirms their ID, if their progress slows, alarms can be automatically raised with senior clinicians who can intervene. Further to avoid these kinds of unfortunate scenarios, AI could be adopted which can scan through the entire hospital resources and predict any possible resource bottleneck in advance avoiding delay in patient care. AI’s ‘Predictive Analytics’ feature is being widely used in most industrial sectors.

AI in Healthcare Startups

Augmedix is bringing AI and Google Glass together to automate scribing, which is now a laborious process of manually taking notes while the doctor is checking patients. Augmedix collects audio, video, and written notes for doctors to make sense of that information.

Another startup, BrainQ, is developing an AI to customize treatment protocols for people who cannot walk because of a stroke, spinal trauma, or brain injuries.

Byteflies is developing a plug-n-play platform for wearables to convert raw data into meaningful data so anyone can use it.

Cytovale’s vision technology looks at how a cell transforms when a patient gets infected with sepsis, helping in the early detection of sepsis.

These developments are indicative of the fact that there is a good demand for AI in healthcare. As such most upcoming healthcare procedures are now being heavily loaded with AI. As AI needs enormous amounts of data for effectiveness, be it diagnostics or procedural, the more data we collect and share across the world, the better it is for AI to make healthcare technologies more productive.

Digital Therapeutics

Today medical knowledge is digitally available online in journals, books, and publications. Patient diagnoses and treatments are now being digitally recorded in all hospitals in the form of electronic medical records (EMR), varied both in ethnicity and demography. Every time a new ailment is treated, it shows up on the patient’s EMR. Every time there is a research breakthrough, the medical information will immediately be available in digital publications. There is constant inflow of digital information and is available in the cloud for the world to take advantage of.

Digital therapeutics is a methodology of diagnosing and treating patients using the data from all the above digital sources. As healthcare data is humongous, it is humanly impossible to digest the big data and provide an accurate diagnosis. This is where AI is contributing.

IBM’s “Watson Health” is a great example that represents an AI driven healthcare algorithm that utilizes host of healthcare data to diagnose ailments. If there is an outbreak of a disease or a medical research breakthrough, AI would learn it instantly as soon as it is digitally available.

The next big thing in healthcare undoubtedly is the evolution of devices that capture human body ‘big data’ in real time. This is where digital therapeutics is headed supplicating both job opportunities and entrepreneurship, all geared towards paring down healthcare costs.

One of the leading digital therapy devices making news today are pills with ingestible embedded sensors that can tell patients and doctors whether a medication has been taken. In November 2017, the FDA approved Abilify MyCite (aripiprazole tablets with sensors). The sensor in the pill sends a message to a wearable patch when the medication is taken. The patch then transmits the information to a mobile app so that patients can track the ingestion of the medication on their smartphone. They can also permit their doctors and nurses to access the information. This is especially great for patients with dementia or mental illness, and surely this will evolve for regular patients who can’t keep track of medication because of busy work schedules. Better patient management improves their health, resulting in reduced healthcare costs.

For certain people with diabetes, the biggest innovation may be Dexcom’s sensor, which displays glucose data on the mobile phone. A Dexcom sensor with a hair-thin wire is placed just under the skin. A transmitter clips to it and sends glucose data via Bluetooth to the Dexcom receiver and then to a smart phone. Even though this device has limitations, with diabetes on the rise, we could expect more investment to go into the monitoring of blood sugar, which will advance the technology behind preventive healthcare.

Glooko offers software for diabetes patients to collect data from Internet-connected insulin pumps and blood glucose meters. Another company ‘Blue Mesa Health’ offers diabetes prevention solution delivered via a wireless scale, Fitbit Flex, and smartphone.

“Asthma Inhalers” are becoming very popular now-adays as they connect to the cloud and captures local airquality data to provide personalized feedback to patients. The Australian company Adherium has come up with a monitoring device for AstraZeneca’s Symbicort aerosol inhaler, dubbed the SmartTouch. Propeller Health in partnership with GlaxoSmithKline has come out with a sensor-enabled inhaler to monitor usage and provide biofeedback. These sensors record the date and time the inhaler is used and transmits the data to an app on the patient’s phone. This history of patient medication data helps physicians make evidence-based decisions and greatly helps healthcare cost-reduction programs.

There are also many GPS enabled smart wearables in the market today like Fitbit, Apple, Garmin and Samsung watches that can track multitude of human activities like jogging and swimming and capture data like heart rate, step count, swimming stroke count, calorie count, etc. Some devices also have sleep trackers that can analyze sleep patterns and recovery. Fitbit is teaming up with Dexcom for glucose monitoring on its smartwatch.

All these sensing devices feed data to AI-driven apps on phones that helps patients track their own health, reducing doctor visits while at the same time doctors use the patient’s data more efficiently, reducing hospital readmissions. All these devices are augmenting the emergence of ‘Preventive Medicine’, yet another booming specialty in healthcare.

The Dark Side of AI

Amidst all the exciting news, there is a dark side to these disruptive technologies. They are forcing doctors to learn new skills to use everchanging techniques. Those who can’t or won’t adapt new technologies might see their career stall. Nursing and other supporting personnel including hospital administration jobs are experiencing similar adaptability challenges.

American Association of Medical Colleges predicts a physician shortfall of about 95,000 over the next decade as more doctors retire early as many of them cannot adapt to new challenging technologies. Physician shortage significantly affects the cost of healthcare. It is not hard to visualize that very soon we will see hospitals with more young doctors who are trained to cope with modern devices and robots.

Fortunately, AI has an answer to this strange scenario. Every one of the healthcare technologies discussed above, when augmented with AI, requires less doctor engagement. We are already seeing these effects in the healthcare industry today. Many new healthcare instruments and procedures are already loaded with AI to help doctors handle them with ease. When new healthcare technologies become more AI centric and less human-dependent, productivity would increase reducing the costs modestly.

Would AI Eventually Replace Doctors?

AI is evolving in healthcare not to replace doctors but to empower them to save more lives. As time progresses, undeniably AI and healthcare will become inseparable. To go from a highly doctor-regimented system to an AI- centric system is a unique journey that all of us would experience in the years to come. This won’t happen overnight. It is a work in progress. It is not easy to fully hand over procedures to a machine, even with the support of AI. Hospitals are ethically and legally obligated to make sure new AI driven procedures are 100 percent safe before they are implemented. For instance, no matter how technologically smart the da Vinci robotic surgery is, it will take a while before it becomes fully doctorless as human lives are at stake. As healthcare is highly risk-centric, the evolution of technology in healthcare will be much slower than other non-healthcare sectors. Nonetheless, that transition has begun.

Conclusion

In this article you may have noticed that healthcare technologies and AI are entirely two different aspects. AI is not a healthcare technology. It is just a computer algorithm that uses healthcare data to make the healthcare technology work more efficiently. Having said that, we could broadly categorize the healthcare technologies into two groups. First group being old existing technologies that are already in the market with little or no AI features. The second group would be the upcoming new technologies. The old technologies will eventually be augmented by AI in their later versions. However, the new upcoming technologies would come with AI loaded, otherwise it is hard to market them. Even these new technologies will constantly undergo updates with latest AI features. In the future years, until all technologies are fully upgraded, we will see a mix of both worlds.

In this article you may also have noticed that IoTs and AI depend on each other. While IoT is a data producer, AI is a data analyzer. While AI needs IoTs to generate quality data, IoTs need AI to consume and process the data. The gist is that the big data that IoTs produce is not of much value without AI. At the same time, AI algorithms are not much value without high quality IoT data. This interdependency is what makes them a powerful duo in digitizing the healthcare system.

AI is undeniably taking the role of a BIG BROTHER in our healthcare system as it can do due diligence of healthcare data, predictive analytics, patient diagnosis, monitoring of procedures, monitoring of inventory, the list is endless.

If you look at all these developments, future healthcare evolution cannot ignore AI, opening a plethora of opportunities both in entrepreneurships and investments. About the author: Ravi Amblee is the author of THE UGLY FIGHT, the award-winning book that deliberates on artificial intelligence as a weapon against climate change. His latest book RISE OF THE DIGITAL WORLD explores India’s opportunities in global digitization. After earning a master’s degree in mechanical engineering, Amblee pursued an IT career that spanned over twenty years while also completing studies on artificial intelligence at MIT, USA. He has contributed articles to several leading Indian publications including the Deccan Herald, the Hindu newspapers and Outlook India magazine. He has a readership of more than seventy thousand on Quora and has answered more than three hundred questions about ways to make technology work for humanity. Ravi lives in Chicago with his family.

Conflict of Interest

None.