Previously we discussed 5 of the 10 ways that Artificial Intelligence is shaking up healthcare. In this second article of the two-part series we will be delving into the second set, as well as offering a set of recommendations on how we can move forward practically with artificial intelligence in our work scenarios.
Before you start, we encourage you to read the part one of the series.
6. Health assistance and medication management
Without nurses there is no healthcare. Throughout my first year as a medical intern I worked around nurses who played a crucial role in my medical education and my understanding of the complexity of roles and relationships within a healthcare setting. But even though the increasing need for nurses and physicians is recognized globally, nursing and medical schools can’t seem to catch up with demand.
That’s where artificial intelligence comes in. Molly, developed by Sense.ly, is the world’s first virtual nurse. Through innovative and avatar-based technology Molly mimics the bedside manner patients need. It plays a crucial role especially in managing patients with chronic illnesses who require long-term management, personalized monitoring and follow-up care.
“But even though the increasing need for nurses and physicians is recognized globally, nursing and medical schools can’t seem to catch up with demand.”
Chronic diseases aren’t the only case scenario where health assistance plays a crucial role. There are situations where patients need to strictly comply with treatment over a shorter time span, such as in Directly Observed Therapy in tuberculosis or for clinical research. The National Institutes of Health recommend the AiCure app, which uses a smartphone’s front camera and AI to visually confirm medication ingestion and thus ensure that patients are adhering to their prescriptions:
Furthermore, there is the incredible story of Neura AI, an Artificial Intelligence startup created as a result of the illness of the founder’s close friend. The founder’s pursuit of answers helped the doctors diagnose a rare form of diabetes. Neura AI is now providing the algorithms for Medisafe, an interesting pill reminder app. The algorithms of Neura AI learn about the patient’s routine, such as what they time they wake up and go to sleep. As a result of this information, and more importantly, Neura’s analytics, Medisafe is now capable of better informing patients about their medications at the right time.
7. Precision Medicine
Genetics and genomics will also be critically affected by Artificial Intelligence, and thus help to foster personalized medicine, which as defined by the National Institutes of Health is "an emerging approach for disease treatment and prevention that takes into account individual variability in genes, environment, and lifestyle for each person."
Deep Genomics has brought together world-leading expertise to create a generation of computation technologies — deep learning — that predict DNA alterations. In his provoking February 2016 TED talk on “How to read the genome and build a human being”, Riccardo Sabatini showed that his team had succeeded in predicting the physical features of an individual using one vial of blood and specific machine learning techniques.
The applications of precision medicine include some of the most prevalent conditions. For example, Verily Life Sciences, a company owned by Alphabet Inc (Google’s mother company), is working on a number of projects focused on Precision Medicine, tackling diseases such as Parkinson’s disease, multiple sclerosis and cardiovascular disorders.
8. Drug discovery and creation, biomarker development and aging research
Drug development takes time and costs money, usually billions of dollars. Using AI, this process can be sped up and made much more cost-effective. Atomwise has taken up the challenge through the use of supercomputers. By launching a virtual search for safe, existing medicines that could be repurposed to treat the Ebola virus, they have been able to find two drugs which have the capability to reduce Ebola infectivity. Using traditional methods, the analysis would have taken months or years, but with Atomwise technology it was completed in one day.
Simulations are also playing an interesting role in drug discovery and creation. In fact, InSilico Medicine, a data analytics company, is on a singular mission to extend healthy longevity and is doing this through the use of artificial intelligence for drug discovery and aging research. The company collaborates with academia, pharmaceutical and cosmetics companies and has licensed more than 827 drug-disease predictions and biomarkers. Their simulations technology reduces the need for animal testing and human clinical trials.
InSilico also believes they can contribute to aging research by looking at pathologically activated pathways specifically related to aging using high-power GPUs.
9. Medical imaging
Medical imaging is also ripe for intervention by innovation. Alphabet Inc. has taken the lead through Verily Life Sciences to develop a machine-learning solution for diabetes-related eye disease through retinal imaging. Given that over 400 million adults suffer from diabetes, this is definitely worth the effort! And retinal imaging is not the only imaging modality being reinvented by Artificial Intelligence.
Machine learning and diagnostic radiology
Zebra Medical Technologies, through their product Zebra Medical Vision, is focused on using advanced machine learning with medical imaging to help diagnose conditions. Their YouTube video below explains what they’re doing in a simple way: “We are teaching computers to detect and diagnose critical medical conditions” through “automated analysis of millions of real-time and retrospective imaging studies.” Just imagine the impact that could have on Population Health Management.
Static and dynamic imaging
In the previous article, we mentioned clinical decision support. Now imagine the possibilities of radiology decision support, where radiologists are assisted with their image interpretation workflow. There are a number of companies working on this, and we will be focusing on two that deal mainly with static images and dynamic images.
Enlitic focuses their value proposition on using deep learning to help doctors interpret images and conduct retrospective analysis more quickly and more accurately through real-time clinical support. There are also startups such as the Butterfly Network that are working on other kinds of dynamic imaging modalities, such as ultrasound.
Medical imaging includes different kinds of images, and among the most important are those related to anatomic pathology. 3Scan took up this challenge with their aim of improving the accuracy and efficiency of anatomic pathology by transforming a manual, analog, and qualitative field into an automated, digitized and quantitative medical science.
10. Public health
Have you heard about Zika? Do you remember the Rio 2016 Olympics? As you might have realized, Zika was well controlled throughout the Olympics, and an epidemiology startup called A.I.M.E played a major role.
The company’s expertise is with vector-borne diseases, which are transmitted by a vector such as a mosquito. A.I.M.E’s platform essentially provides its users with the exact geolocation and date of the next infectious disease outbreak, three months in advance - quickly and intuitively with an 86.37% average predictive model accuracy. Along with the prediction, the company incorporates a fully customizable analytics platform to make sense of the user’s public health data, providing time charts, historic mapping of diseases, rumor reports from social media, and more. Given that the mosquito is responsible for millions of deaths per year, any life-saving measure is worth special attention.
What will it take to get us there?
Despite the gloomy outlook that thought leaders such as Elon Musk and Stephen Hawking expressed in regard to Artificial Intelligence in their open letter, there is huge potential for AI to make a significant, positive impact in healthcare. Infermedica believes that there are four pillars we need to work on to achieve that:
- Information Infrastructure, Innovation and Research – Healthcare systems worldwide need to adapt their systems (a significant number still use paper) to support Artificial Intelligence through the creation of electronic and interoperable systems. All of this needs to take place in an environment which is open to collaborating with digital health entrepreneurs.
- Policy and Legal Framework – Policy, strategies and a robust legal framework are required for the ethical use of artificial intelligence within the healthcare setting without compromising patient data privacy.
- Capacity Building – The increasing use of artificial intelligence requires individuals who are able to direct projects of implementation and ensure that these projects are brought to fruition.
- Education – Patients need to understand that artificial intelligence is to play a role in their care. Healthcare professionals need to be provided with basic training on how AI works and how it supports rather than replaces them in their day-to-day work. Companies need to focus on educating the general public about the potential advantages and risks of using AI in medicine.
“The potential of AI to accelerate the development of healthcare to the next level will ultimately benefit the very heart of healthcare, the patients.”
The potential of AI to accelerate the development of healthcare to the next level will ultimately benefit the very heart of healthcare, the patients. This technological transformation will affect our day-to-day lives as we will be able to spend more time with the ones we truly cherish after being supported by AI-powered healthcare.
Let’s follow up with a discussion on Infermedica’s Twitter Account! Make sure you include the hashtag #healthai and my Twitter handle @stefanbuttigieg.