If one of the arteries in the human brain is blocked it is called having a stroke. If this state lasts longer than 5 minutes, it will either lead to death or to disabilities. But what if people wore smart sensors. It would realise the situation before it happens and would call the ambulance before the stroke appears. Artificial intelligence technology is not only used in our mobile phones, cars, smart homes. One of the most important areas of use is health services. How to use artificial intelligence and health data is among the most important developments of recent times. The most important of these developments is that we can save millions of lives with artificial intelligence thanks to early diagnosis, disease prevention, and in the last instance cure. In the first part of this article, several advancements of artificial intelligence conducting surgery and medical diagnoses using sophisticated algorithms will be explained. In the second part, we are describing the potential of future health applications using wearable technologies to collect huge amounts of individual data to predict upcoming diseases and save lives.
AI in the hospital
Say hello to Dr Robot
The first breakthrough to be named in the field of health care and artificial intelligence is automated surgery. Introduced in 2016 as a robot that can perform surgery more successfully than humans, the robot named Star can now perform unmanned surgeries. The surgery, which was performed with Star and not directed by any specialist, is seen as the herald of a new era in robot technology. The patients in the operation, in which a robot performed the surgery, were four pigs who underwent the procedure in which part of the small intestine is removed to treat bowel cancer in humans. The operation required approximately 20 stitches spaced 3mm apart. Complications that could arise are an incorrectly placed suture which can cause leakage and serious complications. In the operation proceeding with the keyhole surgery method; The instruments are inserted through the approximately 1 cm to 3 cm long incision made in the patient’s abdomen. At this point, the range of motion decreases the patient has limited vision, and the bowel moves with each breath that the patient takes. But during operation, the robot is said to be excellent. An artificial intelligence algorithm helped synchronize its movements with the rise and fall of the pig’s breathing which is far superior to what humans are capable of, while a newly developed camera system known as a structured light endoscope gave it depth perception. At the end of the operation, the artificially intelligent robot surgeon sealed the pig than human surgeons can do and finished the operation.
Machine learning against cancer
However Artificial intelligence (AI) is not only emerging in surgery but also in oncology. In the field of oncology, new AI algorithms are able to diagnose cancers, generate chemotherapy treatment recommendations that are adjusted in real-time based on patient responses and even detect tumours. For example, Google created a software with various healthcare functions. It is successful in predicting a patient’s time in hospital, assessing the likelihood of readmission and mortality. The AI created by google was very successful in tests. Tumours could be detected with an accuracy of 92.4%. This surpassed the highest result previously achieved with another AI method with 82.7%. In the manual measurements of the pathologists, this rate was 73.2%. Despite this high accuracy in cancer detection, google states that they need larger data sets to finish the product.
Another similar product is IBM Watson. IBM’s AI technology, Watson, is now used in more than 230 hospitals around the world, with 55 hospitals using Watson to help diagnose various types of cancer. Although IBM reports that Watson can identify tumors with 93% accuracy, most physicians are unsure of the consistency of this data. Although many of these doctors believed that Watson used real patient data to produce these recommendations, they realized that Watson was using hypothetical data and simply relaying recommendations from other doctors rather than synthesizing its own results from large datasets.
However, in the treatment of cancer not only the best choice of the best available medicine is important but also to keep the side effects as small as possible and to detect the cancer as soon as possible when it starts in the human body. To mitigate the debilitating side effects of radiation and chemotherapy, a team at MIT developed a model that could reduce the toxicity of cancer treatments through a machine learning algorithm. A simulation study of 50 patients was conducted showing that the algorithm can reduce the toxicity of these cancer treatments by up to 50% while maintaining the same effect on tumour shrinkage. The last example that has to be added is that the Harvard Medical School used deep learning to reduce the error rate of human diagnosis by as much as 85%. Thanks to this deep learning mechanism, the computer was able to detect cancer with 92% sensitivity.
AI in the home
The smart toilet
Another issue is in which artificial intelligence has been applied for health care are toilets. Going to the toilet we end up giving important pieces of tissue that can be analysed for human health and artificial intelligence is being used to analyze it. In medicine, the analysis of faeces is been done for decades however thanks to artificial intelligence this can be done with a device that can be attached to the toilet to detect the symptoms of various diseases in stool and urine. Fitting inside the toilet bowl, the device uses cameras, and motion detection technology, and test strips to analyze faeces and urine and send data to a secure cloud server. The technology uses fingerprint scanning and anus images to distinguish users. The device analyzes the basic biochemical composition of faeces and also the users urine samples and makes the physical analysis evaluation based on the Bristol Stool Chart. A study showed that the smart toilet is able to identify about 10 possible diseases which also include cancer. However, despite being so versatile, larger studies are needed to clarify the accuracy of the medical assessments.
Robot nurses on their way
In the pandemic starting 2019, the world has seen a shortage of healthcare personnel. And health care personnel is another field in which artificial intelligence emerges. One step in this field is done by the creation of the healthcare robot Grace. Grace was created in 2021 to respond to the need that emerged with the Corona pandemic. It is supposed to care for isolated patients but also for older people who need care in old peoples homes. It is described as a social robot, which means that it is programmed to help communicate, give therapy sessions and provide social interaction even in difficult situations. Developed in collaboration with artificial intelligence company Singularity Studio and Hanson Robotics, Grace is equipped with a camera and temperature-sensitive sensors to measure temperature and heart rate. The data the robot collects can be used to help experts diagnose diseases and recommend treatments. The mass production and the worldwide application however did still not happen.
Wearable technologies and health data
Wearables, what’s that?
Wearables are small electronic and mobile devices, or computers with wireless communication capabilities incorporated into gadgets, accessories, or clothes, which can be worn on the human body. The popularity of these types of devices has been increasing over the last years such as smartglasses, smartrings and smartwatches from big tech companies. Wearables come in different shapes and sizes, cost and capabilities, and target different consumer profiles. The commonality between the different types of wearables is their ability to collect, transmit or even analyse data collected from the user using various sensors and computational parts. Examples of metrics provided by these devices are respiratory status, blood pressure, heart rate, blood oxygen levels, skin conductance and more. Wearables seem to hold a big promise to provide sophisticated health monitoring at a large scale, and they might even be used to prevent diseases before they even arrive. But what does science say about these technologies, what applications have shown positive results in the lab, can they be used to save lives, and are there any issues to be addressed?
The science and applications of wearable technologies
Research has shown that wearable technologies can be used in many medical applications such as health monitoring, automatic detection of diseases, and prevention of diseases such as type 2 diabetes or seizures. This is promising news for the future since you might not need to attend physically at the doctor’s office anymore for the yearly checkup. Instead, you send the data collected from your various wearable sensors for the doctor to analyse remotely. Aside from medical purposes, these technologies have many applications in other domains such as industry, eyewear or fitness. There are thousands of products on the market and there’s been a big increase in the popularity of wrist-worn products such as wristbands and smartwatches (e.g., Apple Watch and Fitbit). Moreover, smartwatch shipments worldwide are forecast to grow in the following years, reaching a staggering amount of over 253 million units sold by 2025. With those numbers in mind, it’s safe to say that wearable technologies could revolutionise general health at a large scale in the future.
“You don’t want to get to know about your risk of having a heart attack by actually having a heart attack”, said Michael Snyder, a professor in genetics at Stanford University. Apart from his research at Stanford, he’s appeared in media several times such as TED-talks and various public interviews. Snyder talks about the promise of a data-driven approach to medicine using wearables to collect huge amounts of information about individuals each day. He believes that in general, more data is always better to paint a full picture of the human health condition. This sure sounds promising, but are there any examples of how wearable technologies can benefit society at a large scale? In 2021, Snyder and colleagues published an article in the top science journal Nature, presenting the results from a study conducted at his lab at Stanford. The aim of the study was to construct a prediction model, trained on sensor data from more than 5000 participants’ smartwatches, to predict an ongoing SARS-CoV-2 infection. Using sensor data from heart rate, sleep and step count, the model was able to predict infection in 80% of the infected individuals. Moreover, these pre-symptomatic signals from the sensors were observed at a median of 3 days before symptom onset. Moreover, the app would also alert participants if a potential infection was detected. This allowed participants to isolate themselves and get tested even before the symptoms arose, which minimised further spreading in society. The algorithm was not perfect, however, as “the algorithm can’t differentiate between someone who’s knocked back a few too many, or someone who’s stressed because of work and someone who’s ill with a virus.”, Snyder said.
Another study was conducted at the University of Waterloo, where so-called smartshirts were used by the participants. Only after using the shirts for four days, the researchers were able to accurately predict health-related benchmarks during daily activities. “In the near future, we believe it will be possible to continuously check your health, even before you realise that you need medical help”, said Thomas Beltrame who led the research. The study investigated the use of wearables and AI to predict individuals’ health outcomes such as the onset of a respiratory or cardiovascular disease. But as Snyder and other researchers that investigates the potential of these technologies, more work is needed before this approach can be rolled out on a larger scale.
With that said, there are already consumer products on the market today which have demonstrated how they can save lives, such as preventing kidney failure or heart attacks. Apple CEO Tim Cook stated his view of the importance of the technology provided by their Apple Watch to USA Today: “I think you’ll be able to look back at some point in the future and Apple’s greatest contribution will have been to people’s health.”
Why aren’t we wearing them?
Besides the issue that not all sensors are very accurate in many consumer products, and that many challenges still exist to refine the algorithms, why aren’t all of us wearing these technologies already? Researchers who have addressed the challenges for these technologies, argue for the importance of making the devices cost-effective. For example, the price of the latest Apple Watch is 399 dollars, which many people cannot afford. Although there are many other cheaper alternatives on the market today, it’s also important that these devices meet user preferences such as being lightweight, having a long battery life and other aesthetics. It has been shown that about half of people who purchase a wearable stop using it, and one-third of them do this before six months. Another issue is that many elderly lack awareness of these wearables, but do show interest in using them if they are aware of their existence. With this in mind, it seems like it’s only a matter of time before everyone is using some of these devices for health purposes, and probably revolutionising health as we see it today.
The privacy issue
Lastly, it’s important to mention a key issue with these technologies, which is a general concern of the usage of big data and AI: privacy and cybersecurity. Scientists and engineers are working hard to make these systems safe for individuals to use. We have seen what can happen when our data is collected and used for other purposes than users initial intentions, such as the Cambridge Analytica scandal. Like many people who use social media to connect to friends and family, these technologies will be used for private health purposes, not to share their data with third parties. Hence, it’s important to take the privacy issues very seriously for this technology to move in the right direction.
This is the future
This article has reviewed the current and future applications of artificial intelligence technology and data-driven approach to healthcare. Among the various applications, there’s autonomous surgery, autonomous cancer diagnostic and treatment prediction, and automated analysis of faeces. Furthermore, it demonstrated the healthcare robot grace and wearable devices that monitor health so humans get no ill in the first place. It is not possible to predict the future but some things may not seem to be too far fetched. One of these is that in the future, humans will not get as ill as they used to be because of the devices that automatically analyse our bodies so that a stroke may lose its deadliness and colon cancer will not be able to spread because it is detected before that in the toilet. Furthermore, it seems that radiologists and surgeons will be the first to perfectionised by artificial intelligence. Because why use humans if machines are better and faster? Will there be a robot that can do both oncology and surgery one day? And what about the other subfields of medicine? These possibilities raise the question of whether all health care will be replaced by artificial intelligence one day.