Prevention instead of therapy, universal tests, and big data-based diagnostics: will there be a place for human doctors among those new technologies?
Present-day medicine is more than just an interaction between doctors and patients that leads to a diagnosis or a treatment. Nowadays, its objectives include the prevention and early detection of progressing diseases, precise diagnostics, and choosing the right therapy that will fit the patients while not affecting their quality of life.
Diagnostics have become much more sophisticated. Medics not only rely on their (or their colleagues') experience but also access recent scientific papers, have telemedical meetings with experts from another part of the world, and get assistance from AI that is capable of analyzing histological sections or MRI scans as good as any expert.
That is not the limit: medical diagnostics will see even more discoveries. In this article, we will focus on its most promising tendencies.
1. Preventive medicine: anticipation is better than treatment
Life expectancy is increasing, mainly because medicine learned to cope with many diseases that were previously considered incurable; even if some of them cannot be fully treated, they turned from fatal to chronic and can be properly managed.
However, longer life expectancy caused other problems, as people now have to experience more diseases than previous generations. Thus, the best strategy in this situation is timely prevention, i.e., detection of risk factors of certain diseases and early diagnostics.
Screenings partially contribute to this strategy because they are supposed to identify people belonging to risk groups and find a disease at early stages. The sooner the disease is detected, the easier, cheaper, and more efficient its treatment is.
Still, patients are hesitant to take a screening or a preventive examination. Studies show that many people, even if they experience any symptoms and realize they have some health issues, do not visit a doctor because they are afraid of hearing bad news. On the other hand, healthy patients often have no intention of seeing a therapist, as they do not have any apparent health concerns.
This problem can be partially solved by checkup programs encompassing several examinations for health status evaluation. This solution does not suit everyone, though, as many consumers would prefer taking a single test to get complete information on their health.
Another solution that will be in high demand involves introducing AI-based technologies to clinical practice. Those technologies will be able to process large amounts of data and make conclusions on patients' health status by analyzing their analysis results or scans, e.g., X-ray images. Yet most AI systems are still under development and unavailable for clinical use.
On the contrary, the liquid biopsy requiring a simple blood test instead of collecting tissue samples is no longer considered improbable and is being widely introduced to clinical practice.
It has become possible to detect several unique DNA sequences associated with specific cancer types in the patient's blood. Those sequences act as a kind of tumor marker that could help make the correct diagnosis.
AI involvement helps significantly simplify and accelerate the identification of tumor patterns in test samples. Those approaches have already been adopted for diagnosing lung cancers, colorectal malignancies, and other cancer pathologies. GRAIL is working on a universal test based on liquid biopsy methods that could detect 50 cancer types.
2. One test to reveal everything
Various research groups are trying to create something similar not only for cancer detection but also for general health assessment. For example, the US company SomaLogic is developing a test to uncover potential health risks and detect the diseases patients already have. All it takes is a blood sample; the meticulous analysis will measure approximately 5,000 various proteins, enabling the specially trained AI to make conclusions on health status and risks of numerous diseases including cardiovascular disorders and type 2 diabetes.
Scientists are fascinated with the prospect of having a single test that could address most, if not all, health problems. Many areas of medicine are actively working on those developments, with prenatal diagnostics demonstrating outstanding progress over the past 10-15 years.
In addition to ultrasound tests and screenings performed during specific weeks of gestation, non-invasive tests have been developed and actively applied. As those tests were becoming more and more popular, they helped reduce the number of invasive interventions because all they require is a pregnant woman's blood sample. That sample is sent to a lab, where medics analyze the fetus' DNA sequence extracted from the mother's blood.
Non-invasive tests have a high possibility of detecting numerous disorders before the child is born and can assess the risks of pathological development with the help of bioinformatics. Such utmost precision became possible thanks to machine learning and the ability to process vast amounts of data; the databases containing multiple patterns enable the software to estimate the probability of certain pathologies.
Even though the present-day tests can detect dozens of disorders with relatively high accuracy, diagnostic methods call for further improvements. For instance, most testing systems can reveal the most common trisomies related to 13, 18, and 21 chromosomes but fail to deliver any tangible results for trisomies caused by other chromosomes. Besides, current tests do not provide sufficient accuracy in the case of multiple pregnancy; even when a high risk of pathologies is detected, it is impossible to identify which of the fetuses is exposed.
3. AI diagnostics
When it comes to modern medical diagnostics, most doctors, scientists, and patients have high hopes for artificial intelligence, machine learning, and big data analysis.
Such opportunities were first mentioned in 1972 when Stanford University specialists developed MYCIN - the system for analyzing causes of blood infections and predicting the most efficient therapy. Although the system has never been used in practice and remained a generalized concept, it proved its effectiveness; in most cases, MYCIN assigned a better therapy than highly qualified specialists. Back then, its clinical application was impossible; there was no Internet access, and the analysis was extremely time-consuming. However, MYICN developers demonstrated the potential of using artificial intelligence for diagnostic purposes.
Nowadays, machine learning and artificial intelligence can boast a broader scope of application: supercomputers and cloud services have made it possible to analyze vast arrays of data, find common patterns, detect discrepancies, as well as make conclusions based on CT and MRI scans, histological samples, and photos of skin neoplasms.
It takes only a few minutes to complete an analysis; even the best experts cannot give any results so fast. A 2015 study claims that the workload of radiologists is so extensive that they have to interpret CT or MRI scans every 3-4 seconds throughout their 8-hour working day. This strain inevitably leads to mistakes related to human factors, while computer programs are devoid of such constraints, and their usage costs less than an average day of a medical specialist.
Promising developments in this area include PAIGE technology for cancer diagnostics. The AI trained on the database of oncopathology images can make a faster and more accurate diagnosis than any expert, as well as detect an anomaly that even highly skilled specialists would have missed.
In 2021, Paige Prostate, one of the systems using this technology, became the first digital pathology solution that got an FDA clearance. Its application helped reduce false negative and false positive results of prostate cancer diagnostics by 70% and 24%, respectively. Another product launched in 2022, Paige Breast Lymph Node, is aimed to detect breast cancer metastases in lymph nodes. This system has not been used in clinical practice yet, but it will be introduced to improve diagnostic results.
Another emerging tendency is exhaled breath biopsy. The development of the so-called breathomics, the scientific discipline that analyzes volatile organic compounds present in the exhaled air, makes it possible to effectively and quickly detect many diseases, including those that are hard to diagnose. The pioneer of this industry, Owlstone Medical company, is currently working on biotechnologies that will help detect several disorders including mesothelioma, a rare cancer type that can progress for decades without any symptoms.
However, computers will not replace human specialists; artificial intelligence will only assist medics who will stay in charge of making the final diagnosis. We must also keep in mind that a human can identify an image defect and differentiate it from actual pathologies, while an AI system is bound by its algorithms and thus is capable of misdiagnosis.
4. No telemedicine without human doctors?
Another trend that has been gaining momentum since the late 20th century is telemedicine, with its development boosted by the Internet and the 2020 COVID-19 pandemic. Within a short time, the whole world had to self-isolate and the medics needed to cope with the heavy burden of saving patients' lives without any effective protocols.
Even though at some point the coronavirus became the most pressing health issue, other illnesses were not going anywhere. Patients kept getting sick; they needed to get a diagnosis or have their therapy adjusted. It turned out that telemedical consulting, during which patients communicate with their doctors remotely via a messenger or application instead of going to the clinic, can solve many problems, simplify the doctor-patient interaction, and save time for every specialist.
Despite its evident advantages, telemedicine has a downside - in most cases, it applies to recurring appointments only. Initial examinations by some specialists, e.g., ophthalmologists or OB-GYNs, are impossible in this format.
Besides, not everyone is ready to trust an AI with making a diagnosis, no matter how many large databases were used for its training. However, special terminals for online consulting recently appeared in China, developed and launched by Ping An company that specializes in digital medical technologies.
Those terminals, slightly bigger than a phone booth, are called "One-minute Clinic". Their "visitors" can get help in the shortest time possible: the AI is capable of online diagnostics and giving recommendations that meet international standards. Patients can even buy medications stored in this "clinic" following all the necessary regulations. If a drug is out of stock, it can be ordered online and collected from the nearest drugstore within an hour.
Still, this technology is not a rule but an exception - most patients want to talk to a real doctor instead of a virtual assistant; few are ready to trust a machine with their health. Experts say telemedicine will continue to improve and develop. This includes adopting AI technologies that will manage routine tasks like collecting medical histories or test results to provide already processed information to the doctors so that they could spend more time with the patient and choose the most efficient therapy.
What should we expect?
The demand for new diagnostic technologies will keep on growing. Patients need fast and innovative solutions that require minimum time and effort to get an accurate diagnosis, while doctors need supporting tools to take over some routine activities. Machine learning will become more sophisticated, supercomputers will process more data, and artificial intelligence will get smarter and make fewer errors; its decisions will be more "human-like" without any decline in speed or accuracy.
About the Author
Rustam Gilfanov is an IT entrepreneur, a co-founder of a large IT company, and a venture partner of the LongeVC Fund.
* This is a contributed article and this content does not necessarily represent the views of sciencetimes.com
![Sat-Nav in Space: Best Route Between Two Worlds Calculated Using 'Knot Theory' [Study]](https://1721181113.rsc.cdn77.org/data/thumbs/full/53194/258/146/50/40/sat-nav-in-space-best-route-between-two-worlds-calculated-using-knot-theory-study.png)
