AI in healthcare: trends and application examples

In this article, we will consider medicine as a system of scientific knowledge and practical activity, the goals of which are to preserve and strengthen human health, prolong human life, and treat and prevent diseases. Under the content side of the term “artificial intelligence,” we will understand technologies based on the training of computer systems designed to replace human actions when performing any processes. This definition will allow us to focus on the practical aspects of AI use cases in healthcare and avoid the philosophical issues that often accompany the discussion of AI in lay circles.

The exponential growth of research is usually accompanied by a constant expansion of the range of problems to be solved. Therefore, we will not pretend to provide an exhaustive picture of AI applications in medicine but will try to outline the most successful or promising areas from our point of view.

AI in diagnostics

IBM estimates that 90% of the data in the healthcare industry is images, and their volume is increasing faster than that of all other medical data. Having started its triumphal march by recognizing images of dogs, cars, and handwritten digits, neural networks have come in handy in processing a variety of visual data.

The capabilities of neural networks are helping to transform the field of radiology, saving time and money for medical organizations. After a medical image is acquired through an MRI, CT scan, ultrasound, or X-ray, the doctor must analyze it for any abnormalities or signs of disease. Interpretation of several imaging studies is required to identify any serious condition.

Once trained using large study datasets, AI-based systems are able to analyze medical images and report on features, such as small tumors, that the human eye might miss. Such systems identify patterns and provide information about the characteristics of any abnormalities, saving physicians time.

In cases where a patient has several images taken over a period of time, artificial intelligence can also analyze the dynamics of the disease. For example, Google conducted an experiment to test its AI-based system: six certified radiologists were asked to examine the images. In those cases, when the diagnosis was made on the basis of a single image, artificial intelligence coped as well or even better than humans. The system was able to diagnose 5% more cases of cancer and reduce false positive verdicts by 11%.

Virtual nursing assistants

According to a recent Accenture report, the use of virtual nursing assistants in the healthcare industry could save $20 billion a year by reducing the time nurses spend on patient care by 20%. Today, computerized assistants already work alongside live nurses in U.S. hospitals, from whom you can get tips, hints, and other information. For example, digital assistant Sally, a smiling woman in a white coat, or nurse Walt. Sally and Walt are animated avatars, virtual personal health coaches from iCare Navigator, an artificial intelligence-based platform designed to interact with and educate patients.

TeleHealth Services, which developed iCare Navigator, says it uses patients’ electronic health records and applies machine learning to build personalized relationships. The app determines when a patient is most receptive to information about their health status and their care can be best managed.

The impetus for the iCare Navigator platform came from research at Boston University School of Medicine, which developed virtual nurses Louise and Elizabeth to explain to patients, for example, when to take their medications. It turned out that 74% of patients preferred to receive guidance at hospital discharge from a virtual nurse rather than a human.

Molly from Sensely is another popular artificial intelligence nurse avatar used by the University of California, San Francisco, and the UK’s National Health Service. Molly asks patients questions about their health, assesses symptoms, and makes recommendations for the most effective treatment based on symptoms.

So, instead of searching the Internet for symptoms that one has discovered in oneself, today, a person can get help from a virtual nurse. Virtual nurses not only provide medical advice on common diseases or ailments but also allow you to make an appointment with a doctor. They are available around the clock and are ready to answer questions in real time. This is one of the major applications of artificial intelligence in healthcare, which is increasingly being used to raise awareness and improve self-management skills in patients with chronic conditions. Thanks to the virtual nurse, the patient will be able to prevent their condition from worsening.

In order to develop such systems, you’ll need to hire AI developers.

Multimodal diagnostic systems

Several trends can be identified in the development of AI, one of which is related to the integration of the types (modalities) of data on which training is performed. For example, for audio-visual speech recognition, the visual description of lip movement is combined with audio input to predict spoken words. Information from sources of different modalities may have different predictive power and noise topology, and some sources may have missing data. The heterogeneity of multimodal data makes model building difficult.

It is important to explore how to represent input data and summarize it in a way that reflects multiple modalities. For example, text is represented by symbols, while audio and visual modalities are represented by signals. In the context of medical applications, all diagnostic information about a patient can be integrated into such multimodal data and processed by an AI system trained to consider both external images of a person and body fragments, as well as test results, MRI and CT images, audio recordings of answers to questions, etc.

All this brings us closer to creating a universal diagnostician, using a comprehensive approach to diagnosing diseases, and reducing the number of visits to various specialist physicians to prescribe effective treatment. If you need to create an app or software for medicine, you can contact a healthcare software development company.

Artificial intelligence-powered health apps

The biggest potential benefit of artificial intelligence is the ability to help people stay awake so that they don’t have to visit the doctor, or at least don’t do so too often. Artificial intelligence and the Internet of Medical Things (IoMT) are already gradually changing the paradigm from “reactive” healthcare to “proactive” healthcare.

The combination of artificial intelligence and IoMT will eventually make connected health monitoring devices smarter. AI and the vast amounts of data generated by IoMT can also be used for diagnosis.

Various AI-based healthy lifestyle apps, such as MyFitnessPal and HealthTap, give people full control over their health and well-being, feedback from their healthcare provider, and recommendations for staying healthy. HealthTap, for example, learns about a patient’s symptoms and how they change over time and coordinates care by sending reminders, providing text responses matched with medical history data guidelines created by physicians, and enabling online consultations via videoconferencing.

Is AI in medicine a breakthrough?

Can the use of AI be called a breakthrough in diagnosis and treatment? In our opinion, a breakthrough has not yet occurred today. Therefore, we would use a quantitative assessment of the development of the technology, for example, the number of successful research projects in this area or the number of publications. If such an indicator grows exponentially, we can speak of rapid progress. From this point of view, we are present in the development of breakthrough diagnostic and treatment technologies.



Leave a Reply

Your email address will not be published. Required fields are marked *