Artificial Intelligence (AI) is enabling healthcare professionals to fight disease with unprecedented accuracy and precision. The use of AI to screen, assess risk, and predict disease is proving to advance the effectiveness of diagnosis, treatment, and management of communicable and non-communicable disease alike. It is expanding the arsenal of powerful tools science and medicine has to combat diseases of all forms.
The Burden of Disease
Technological advancements come at a time when incidences of disease are rising due to changes in lifestyle and the environment. The World Health Organization and the Center for Disease Control report sobering statistics for prolific diseases with high mortality rates.
- Cardiovascular disease is the #1 cause of death in the US and globally (CDC, WHO)
- Cancer is the second leading cause of death globally and in the US (WHO, CDC)
- Globally, nearly 1 in 6 deaths is due to cancer (WHO)
- The global financial burden of cancer is estimated at US $1.16 trillion (WHO)
- In 2015, the US alone reported 1.6 million new cases of cancer (CDC)
- Communicable diseases are the top two leading causes of death in low-income countries (WHO)
- Diabetes accounts for 1.6 million deaths, annually and globally (WHO)
As trends indicate that these numbers will rise, artificial intelligence will continue to play a central role in managing both the public health and economic impact of disease.
The Impact of Artificial Intelligence
Tech giants like Google and Microsoft, and an array of startups, have invested billions of dollars to develop AI technology in the fight against disease. They are developing smart models that allow clinicians to analyze patient data and identify patterns of disease behavior at stages undetectable by traditional methods. These advancements are possible with machine learning (ML), an integral field of artificial intelligence. ML uses statistical analysis to derive probability based inferences, enabling computers to make intelligent disease prediction and detection with superb accuracy. Machine learning has matured to the point where tech companies have developed their own platforms and tools, allowing developers worldwide to take advantage of AI, making it ubiquitous.
Alphabet’s Verily Life Sciences with Nikon is applying deep learning models to detect Diabetic Retinopathy and Diabetic Macular Edema, two leading causes of blindness among adults, much earlier than what conventional methods permit today.
In some cases, AI has proven to be better than humans at certain tasks that involve detailed analysis. Zebra Medical Vision developed an algorithm that accurately detects breast cancer 92% of the time, representing a 10% greater accuracy over the average radiologist using computer-aided detection software. Similarly, a Chinese AI system called BioMind correctly diagnosed brain tumors in 87% of 225 cases in 15 minutes.
A team of researchers at Google has developed a model that uses pattern-recognition techniques on retinal images to predict the risk of a patient experiencing a major cardiovascular event or stroke in the next 5 years from a simple retinal image. The prediction is accurate about 70% of the time, the same rate as conventional methods that require measuring a patient’s cholesterol levels. The newer technique isn’t just less intrusive because it doesn’t require a blood draw, the team believes it can be further perfected by simply increasing the size of the sample data-set that the model was trained with.
Artificial Intelligence Enables Precision Medicine
A startup in San Francisco called Freenome uses artificial intelligence to identify cancer-related proteins among billions of circulating cell-free biomarkers in the bloodstream. This emergent method, also known as a liquid biopsy, has the potential to make a radical impact on cancer survival rates. With artificial intelligence, doctors are able to detect cancer at its earliest stages, sometimes prior to the formation of tumors, helping them prescribe the most precise treatments.
With the right data, artificial intelligence is highly effective at predicting when and where disease outbreaks are likely to occur, helping public health officials prepare for outbreaks well in advance and take preventative measures. Artificial Intelligence in Medical Epidemiology (AIME), a startup at Singularity University, has developed a model that uses 11 different variables to predict outbreaks of dengue fever with a stunning accuracy of 87%. The model accounts for factors such as weather, constructions and dengue death rates to specify a geographic radius of 400m with such impressive accuracy. Such models allow professionals to look at preventative approaches to diseases, revolutionizing the field of medical epidemiology, protecting millions of people and saving billions of dollars.
The Future of Artificial Intelligence in Healthcare
Though still in its infancy, AI shows great potential in helping medicine fight disease. But there is also skepticism. Its performance and output depend on reliable and sizable datasets, protecting consumer privacy, data security, and the fate of the role of doctors. Google’s DeepMind, its AI subsidiary based out of London, England, ran into trouble when the UK’s Information Commission (IC) ruled that its patient data-sharing deal with the Royal Free NHS Foundation broke privacy laws. There is uncertainty over the legality of collecting, sharing, and using medical data as well as concerns about patient privacy. And some are beginning to advocate for regulation to protect consumers. Furthermore, ML models are only as good as the datasets they are trained with and the lack of good training data could make even complex models inaccurate and ineffective.
Even so, investors are pouring millions of dollars into artificial intelligence and healthcare continues to expand its use to help clinicians fight disease and improve patient care.