This year WHO issued its first
It is projected that by 2025, the AI healthcare market will increase to over 28 billion U.S. dollars. That is a positive dynamic, considering the market of 1,4 billion in 2017. But how exactly can medical services benefit from intelligent systems? And what are the main limitations of AI in healthcare?
Let’s find out.
From a historical perspective, AI has been around for quite a while.
It was first described in 1950. Yet, it wasn’t until the early 2000s when limitations were overcome by the advent of deep learning and AI was widely accepted.
By 2017, AI in medicine had transformed into the main industrial application of AI in terms of aggregate
And finally, the global artificial intelligence in healthcare market size was estimated at $6.7 billion in 2020. But what caused the sudden spike in AI investments?
Here’s what has fast-forwarded the growth of the AI market in medicine:
Hence, we can assume that intelligent systems were recognized as a new boundary for medical services. The current state of the AI market in healthcare demonstrates encouraging numbers:
Not a pipe dream anymore, I guess.
From preventative care through enhanced medical decision-making to better patient outcomes, AI is becoming a part of our healthcare ecosystem. Let’s have a look at the most prominent use cases of artificial intelligence in medicine.
Prescreening AI tools can add value even before patients visit medical facilities. Thus, intelligent algorithms can assist patients in self-monitoring and patients-at-home isolation, so that there is no rush and panic at clinics and hospitals.
Today, prescreening boils down to standard questionnaires or online symptom checkers like WebMD. According to BMJ, these traditional pre-screening methods have a
On the contrary, extended AI-based functionality like voice recognition can ensure a more user-friendly experience than typical pre-screening questionnaires. Through machine learning, AI may also discover new patterns and detect patients' symptoms more correctly.
A potential patient could describe symptoms through speech or text, just as they would to a doctor, thanks to natural language processing (where computers recognize text patterns). The patient and AI bot will exchange questions, replicating a real-life discussion between a doctor and a patient.
Also, predictive analytics in healthcare helps to reduce patient admissions. In particular, it reduces preventable readmissions and predicts the possibility of developing a heart failure.
According to the European Radiology Experimental’s poll, more than halfof global healthcare leaders project the impact of AI in monitoring and diagnosis to increase significantly. And there’s some good rationale behind it.
Over the last few years, artificial intelligence has gained momentum within the continuum of medical imaging and diagnostics, thus promoting higher accuracy for medical researchers and doctors. AI-enabled diagnostics eliminates human error as well as shortens lab time and medical costs.
In particular, image recognition has the power to innovate medical diagnostics. Thus, medical imaging is the field that benefits the most from the adoption of smart systems. As such, medical imaging includes numerous radiological techniques. These may include:
AI can be used to discern advanced patterns in imaging data and yield quantitative analysis of radiographic characteristics. Thus, an AI-based technology developed by Lunit was employed in a
Tumor delineation and treatment assessment are other widespread AI applications in radiation oncology.
Traditional drug discovery is a resource-consuming process that can go on for years. Thus, taking a drug from discovery to market can take up around 10 to 12 yearsand incur costs of over €2B.
Luckily, artificial intelligence has gained traction in the drug research industry to ease the strain on the healthcare industry. According to MarketsandMarkets, AI in drug discovery accounts for a global market value of $1.434 billionby 2024, up from $259 million in 2019.
Some of the flashy demonstrations of AI in biomedical research include the AlphaFold programby Google’s AI subsidiary, DeepMind. It is said to yield computational predictions of protein structure that approach the quality of those provided by gold-standard experimental techniques such as X-ray crystallography.
In complex drug discovery, intelligent algorithms hold great potential to accelerate researchprocesses and make them more cost-effective, while also reducing the time a new drug requires to reach the patient. This superpower of artificial intelligence is attributed to its ability to comb through large datasets. The latter can include everything from clinical studies and scientific literature.
The algorithm then uncovers hidden patterns and relationships in data sets within mere seconds. Can manual efforts be on par with this speed? Sadly, no.
Once the protagonists of science fiction, robots have become a reality thanks to advanced AI capabilities. Today, robotic surgeries assist doctors in performing complex procedures with more precision and flexibility. For example, some robots guide surgeons when placing screws during spinal surgery.
Others simulate training and shadow surgeons during procedures that allow professionals to hone up their skills. However, the latter use case sparks up debates as robotics is cited as the
Nevertheless, robotic surgery is an unseen generation of medical manipulation that aims to combine the strengths of technology and the human mind. Imagine a robot whose movements are controlled by a human. Isn’t it the dream of all sci-fi writers of the past?
The proliferation of 5G networks is also carving out new possibilities for treatments using surgical robots. Thus, 5G facilitates remote surgery by instantly transmitting ultrahigh-resolution medical images of massive data size without delaying control signals. 
This year, Kobe University,
In 2020, Microsoft co-founder Bill Gates statedthat advances in artificial intelligence and gene editing could accelerate the improvements of gene-based editing technologies. While we haven’t reached the equilibrium point in genome editing yet, the use of AI in this field bode well for identifying harmful genes and treating disease.
Artificial neural networks are capable of identifying and revealing patterns in massive amounts of genetic data, thus detecting groups and sequences of genes associated with particular diseases.
Also, machines can compute models to
Moreover, machine learning algorithms are also helpful in pinpointing where the alteration must be made and how to ensure the
Currently, artificial intelligence is used to boost the accuracy of a gene-editing technique called CRISPR-Cas. In particular, machine learning
Although delivering clinical impact with AI is certainly promising, there are some clouds in the silver lining. Insufficient medical data is one of the challenges that await healthcare providers on the way to automation. Fragmented and unorganized health data haunts medicine and hampers effective training of AI algorithms.
Healthcare data is often siloed in a wide range of medical imaging archival systems, EHRs, insurance databases, and others. Hence, this patchwork of information is hard to tie together.
However, unified data formats like
Moreover, a complex web of ingrained economic factors and ethical pitfalls also slow down the widespread adoption of healthcare automation. And let’s not forget legal barriers like the absence of AI standards in healthcare and the inability to be used in resource-poor settings.
The possibilities of artificial intelligence stretch across industries. Healthcare is among those fields that are benefiting the most from increased automation and hands-free processes.
Artificial intelligence has the potential to fundamentally change the entire world of medicine. It can transform the diagnostic system, facilitate the development of new drugs, improve the quality of medical services, and reduce costs. However, since we haven’t overcome a multitude of AI limitations yet, the greatest disruption is still ahead (but we’re certainly moving in the right direction).
 Ting DSW, Carin L, Dzau V, Wong TY. Digital technology and COVID-19. Nat Med. 2020 Apr 27;26(4):459–461. doi: 10.1038/s41591-020-0824-5.
 Shameer K, Johnson KW, Yahi A, et al. PREDICTIVE MODELING OF HOSPITAL READMISSION RATES USING ELECTRONIC MEDICAL RECORD-WIDE MACHINE LEARNING: A CASE-STUDY USING MOUNT SINAI HEART FAILURE COHORT. Pac Symp Biocomput. 2017;22:276-287. doi:10.1142/9789813207813_0027
 Hu, Xiaobang & Ohnmeiss, Donna & Lieberman, Isador. (2012). Robotic-assisted pedicle screw placement: Lessons learned from the first 102 patients. European spine journal : official publication of the European Spine Society, the European Spinal Deformity Society, and the European Section of the Cervical Spine Research Society. 22. 10.1007/s00586-012-2499-1.
 Memos, Vasileios & Minopoulos, Georgios & Psannis, Kostas. (2019). The Impact of IoT and 5G Technology in Telesurgery: Benefits & Limitations.