Technological advances like data mining and machine learning promise important applications in the future of health, from improvements in pharmaceutical research to the automatic diagnosis of illnesses.
By Anna Marmyleva
In science fiction films like The Host (2013), we see inhabitants of other planets being cured of serious diseases in days, even seconds: futuristic machines scan their bodies and administer blood tests with minute devices, immediately obtaining diagnoses and treatments. This appears to be magic yet is closer to being possible thanks to big data and artificial intelligence (AI), which ceaselessly create opportunity in the medical field.
In recent years, international healthcare systems have digitalized more than ever and patients provide increasingly more data thanks to the use of smartphones and wearables. AI, thanks to its capacity to examine, structure, and analyze huge databases, allows the discovery of complex patterns and the development of predictive algorithms for disease research, development of new pharmaceuticals, and modeling of clinical trials. Thus, big data and AI are also proving to be useful tools in researching the Covid-19 pandemic.
A slow yet undenniable evolution
Artificial intelligence, born as a concept at a 1956 Dartmouth College conference, was used for the first time in the health sector in the 1970s with MYCIN, an expert system aimed at diagnosing infectious diseases in blood. This pioneering system reasoned, communicated with the user in natural language, and recommended medication in a personalized manner for each patient.
However, the digital transformation and use of new technologies in the healthcare industry has advanced “more slowly than in other sectors,” says Xavier Contijoch, director of Opinno Barcelona.
“Although there have been digital initiatives for a long time, the sector had never regarded itself as being in a place to change. On the one hand, because the industry has a very rigid demand and is highly regulated, and, on the other hand, because the healthcare system is very complex and depends on public policy, which makes a digital transformation very difficult, says Contijoch, who has worked with big pharmaceutical and insurance companies in Spain.
Photo: The digital transformation of the health sector will begin with data exploitation. Credit: Pexels.
Nonetheless, “as technology has evolved, innovation has been incorporated little by little,” adds Contijoch, who explains how this has given way to applications of big data and AI that are increasingly broad and complex.
More than half a century after the first industry use of these technologies, the global healthcare market linked to patient data and AI is expected to reach a value over 19 billion dollars by 2025, according to projections by Tractica.
Diagnosis and treatment
Big data and AI have already begun to bear fruit in the diagnosis of pathologies and patient treatment. Among the latest advances in the diagnostic space is a rapid new test to detect cancer developed by scientists at the National University of Science and Technology in Moscow, Russia.
Thanks to the sophistication of the algorithms behind this immunochromatographic test, it is possible to detect any type of prostate cancer in only ten minutes and with a high probability of detecting the disease in an early stage. Moreover, this same technology enables the creation of deep learning algorithms for the detection of cancer with a system so quick that it could process the radiographies of the entire population of the Russian capital (12 million people) in 30 seconds, according to the researchers’ assessment.
AI to understand medical data
One of the fields of knowledge in artificial intelligence that is advancing very quickly is natural language processing (NLP). According to a 2019 report by the Roche Institute Foundation on the role of data in healthcare, these technique not only facilitate the integration of data, but also the classification of patients in different groups thanks to its capacity to automatically transform text into structured information.
Natural language processing can accelerate decision-making in diagnosis and treatment and improve patient care.
The application of NLP to the healthcare context is fundamental since it helps overcome one of the principal barriers to the efficient use of data in the health sector: the heterogeneity of unstructured data that is stored in different formats. NLP can accelerate the decision-making process for diagnosis and treatment and improve patient care.
A good example of this is Amazon Comprehend Medical, a natural language processing service developed by Amazon Web Services (AWS) that extracts relevant medical information from doctors’ notes, clinical trial reports, and patients’ medical histories.
Furthermore, NLP helps to link detected information with international medical classification systems like ICD-10-CM (International Classification of Diseases, Tenth Revision, Clinical Modification) and RxNorm, which are used in the industry for the homogenization of terminology. This functionality allows for a much simpler use of collected information in the medical practice.
Intelligent pharmaceutical research
The use of AI in the health sector already enables pathology diagnosis and treatment prescription based on a patient’s current data, but it also facilitates the prediction of the functionality of medication, which helps launch new drugs.
One of the most prominent trends right now is “the implementation of data analytical tools in clinical laboratories to accelerate the process of clinical validation of new pharmaceuticals,” says Contijoch. According to this expert, these tools allow for the anticipation of failure of any component of a new drug and for “deciding which molecule to invest resources in.” In this way, when a new medication’s cost of production is high, “having platforms that combine all of the data and produces conclusions generates significant savings.”
Photo: Artificial intelligence can be used to accelerate the validation of pharmaceuticals. Credit: Anna Shvets via Pexels.
AI against the novel coronavirus
The predictive capacity of AI is especially important in extraordinary circumstances like the novel coronavirus pandemic, since it enables considerable risk mitigation for society and improves opportunities to stop the spread of the pandemic. The intelligent and centralized use of patient data and the development of new analytical algorithms is now mandatory in the age of Covid-19. In search of new, innovative ideas that will make optimal use of the data, initiatives like Beat the Vid, Opinno’s open innovation program, have been launched. “With the coronavirus and the maturity of technologies in the healthcare sector, the digital transformation in health is drastically accelerating,” says Contijoch.
Photo: Artifical intelligence is also contributing to the fight against the coronavirus pandemic. Credit: Pexels.
Thanks to these two factors––the vast quantity of centralized data and a powerful AI algorithm––Maccabi Healthcare Services, one of the Israel’s largest healthcare organizations, is identifying people who are at greater risk of suffering grave complications from coronavirus.
Once identified, these people undergo rapid diagnostic testing. According to the MIT Technology Review, by early May this technology had already identified 2 percent of the group with the highest risk, or 40,000 people. The algorithm also indicated the level of treatment that each patient should receive in the event of contracting the disease.
An analytical and technological view on the future of health
As the 2019 Roche Institute Foundation stated, artificial intelligence represents a paradigm shift in healthcare. Recent breakthroughs also indicate that smart technologies could be used to predict future diseases through an analysis of the intersection of environmental and consumption data.
“In the short term, AI technologies are only going to be used to augment, not replace, the operational intelligence of a physician.”
However, since the healthcare sector directly affects people’s health, much precaution is needed. “In the short term, AI technologies are only going to be used to augment the operational intelligence of a physician, not to replace it,” says Contijoch. He adds: “It is insufficient to have tools or drugs that kind of work, they must work to perfection.”
On the other hand, the sector’s rigid regulation will mean that the deployment of these technologies will generate more risks associated to the use of patients’ sensitive personal data. More sophisticated security systems will be necessary. Artificial intelligence could itself be a tool for the monitoring and detection of security breaches, helping to foster greater confidence among society and accelerating the digital transformation in this complex sector.
