Ordering chaos: artificial intelligence in medicine

Artificial intelligence, algorithms, computing power – it sounds like a sci-fi movie. Such tools, however, are entering into new aspects of our lives, including medicine. We are usually afraid of what we don’t know. Should we fear that artificial intelligence will replace the doctor?

Artificial intelligence (AI) is a term coined in 1956 by American computer scientist John McCarthy. Refers to the ability of the system to correctly analyze data. Over the past two decades it has been increasingly used, also in medicine. She helped, among other things, sequence the human genome.

In 1990, the US Department of Energy and the US National Institutes of Health (NIH) decided to allocate $3 billion to the Human Genome Project. The decision was based on 15 years of reading the human genome. The success was announced earlier, after 13 years, on April 14, 2003 to be exact. A document was then published indicating that the sequencing of the human genome was completed. However, it took a long time.

Technology has evolved so much since this groundbreaking event that just a few months ago, in January 2022, a new world record was set, i.e. the human genome was sequenced in just 5 hours. and 2 minutes: from sample collection to delivery of the result to the patient.

Why genome sequencing? Scientists hope, not without reason, that it is a chance to cure many diseases – those caused by genetic mutations. Artificial intelligence is one of the tools that helps in these discoveries.

AI in genetics

Why is AI only now being used in medicine? After all, AI algorithms weren’t invented 10 years ago. We have known them for several decades.

– Yes, but for decades we only knew them theoretically. For practical use, one simple element was missing, namely supercomputers. There was no such computing power to put theory into practice – said Dr. Paula Dobosz from the Department of Genetics and Genomics at the Central Clinical Hospital of the Ministry of Interior and Administration in Warsaw at the conference “Before it is too late, prophylaxis and modern methods diagnostics are a prerequisite for effective therapy”.

– A patient’s genome in such a “raw” form is at least 60 GB, so if we wanted to process such data on an ordinary laptop, it simply cannot be done. And that’s why we’re just starting to introduce artificial intelligence into our daily practice,” she added.

Are we afraid of artificial intelligence?

According to studies conducted on more than 20,000 people from 27 countries, including Poland, between April 10 and May 3, 2019, it results that up to 41 percent. of respondents fear that artificial intelligence could one day replace us.

– On the other hand, those of the respondents who understood what artificial intelligence is and how it can help us in life, see no reason to worry. There were 27 percent of these people in the study. Up to 32 percent had no opinion on this – said Dr. Paula Dobosz.

Artificial intelligence in medicine is a faster diagnosis and support for doctors, but not their replacement. According to Dr. Dobosz, we do not have to fear that the computer decides how to treat us, because artificial intelligence is associated, for example, with Google Maps, which works on a similar principle, or with a smart vacuum cleaner that maps , “learns” our apartment system and then knows how to clean.

The use of artificial intelligence in medicine is:

  • Organization of medical records;
  • Modeling and prediction of treatment response;
  • Analyze voice and speech, which can be useful, for example, in the early detection of neurodegenerative diseases;
  • Imaging (computed tomography, magnetic resonance imaging, X-rays) and on the basis of these tests, diagnosing, for example, tumor foci or strokes;
  • Pathomorphology: analysis of digitized IHC images (quantification of staining intensity);
  • Cardiology: monitoring the functions of automatic defibrillator devices;
  • Systems supporting patient condition monitoring (applications), eg risk of heart attack, recurrence of neoplastic disease, hypoglycaemia;
  • Analysis of genetic data.

Dr. Paula Dobosz reminds that artificial intelligence must encourage, standardize and be an additional source of information for the doctor.

– He will support the doctor, but the algorithm will not make the decision for us, it is only supposed to support him. It will analyze data in real time, provide information and monitor the patient’s condition, especially outside the hospital – she stressed.

According to her, artificial intelligence will increase the availability and safety of medical procedures.

What can we read about our genome?

The sequencing of the human genome and the use of artificial intelligence in medicine are already used in the treatment of cancer.

– Variants associated with disease risk can be identified. Perform molecular analysis of neoplastic genome and patient genome and customize modern therapeutic therapy. Each cancer is a disease of our genome. Additionally, we can identify variants that affect treatment response or non-response and apply targeted therapies and monitor treatment progress, explained Dr. Paula Dobosz.

Out of curiosity, she mentioned the so-called mutation signatures. – From the patient’s genome, you can find out that someone is 100%. he smoked cigarettes, although the patient denies it. The substances contained in cigarettes are the most mutagenic. They still leave a mark on our genome, so for the rest of our lives we will be able to identify it using genomic mutation signatures, explained Dr. Paula Dobosz.

The first country to decide to test whole genome sequencing in cancer patients was England. The pilot program involved several types of cancer, and after its completion it turned out that up to 36%. the effectiveness of treatments based on genomic data has increased. Currently, England is expanding the program to include more cancers.

– As a result, they save time and money as well as a higher patient survival rate – she pointed out.

Artificial intelligence in imaging studies

Since 1895, when the discovery of X-rays made a milestone in medicine, imaging research has undergone a revolution. Then the irradiation lasted 20 minutes – now it is almost unimaginable and unacceptable, because at this time the patient absorbed a huge dose of X-rays. Imaging diagnostics was improved and it was possible to develop ultrasound examinations using ultrasonic waves. Today, medicine has computer tomographs, magnetic resonance imaging that uses a magnetic field, which can image the functions of organs and organs in the human body, or scan the whole body to find pathological changes.

– Each of these methods has advantages and disadvantages. In each of the imaging diagnostic methods, there are disturbances, artifacts, incompletely imaged areas, that is, the examination did not accurately describe the examined area, and the radiologist did not not been able to accurately assess the changes. Sometimes the disease is so early that it is difficult to assess. This is where artificial intelligence comes to the rescue, explained Dr. Maciej Bobowicz from the 2nd Department of Radiology at Gdańsk Medical University.

As an example, he gave a mammogram, which involves detecting neoplastic changes very early, and a radiologist may have difficulty evaluating the image for several very simple reasons:

  • Very subtle traits of malevolence;
  • A slow change in appearance from normal to cancerous;
  • Dense glandular tissue masking lesions;
  • Inconspicuous neoplastic changes;
  • Limited areas of asymmetry;
  • Appearance evoking a benign lesion;
  • Lesions located outside the imaging area;
  • The human factor, ie the fatigue of the diagnostician.

This is where artificial intelligence comes in handy.

– We are currently implementing the grant “European Radiological Research Platform integrating clinical data based on next-generation artificial intelligence for targeted medicine in oncology”, hoping that artificial intelligence will help in diagnosis, but it does not help us will certainly not replace – said Dr. Maciej Bobowicz.

According to him, algorithms, applications, or whatever we call them “can” mark changes that don’t give symptoms, as well as picture changes that aren’t entirely clear.

– When we manage to detect such a change with the help of artificial intelligence, we are able to provide its characteristics, i.e. create a classification and determine to which group a given change may belong . Also, we can determine the magnitude of the change. Artificial intelligence can use classifiers of the type of malignant and benign changes. We are able to define the types and subtypes of cancers we treat, as well as the stages and groups at risk, which translates into the quality of treatment. We are able to make the correct diagnosis more quickly, underlined Dr. Maciej Bobowicz.

The benefits of artificial intelligence in the treatment of breast cancer:

  • Detect minor changes;
  • Early detection of the disease;
  • Ability to analyze complex and more complex images;
  • Reduce the number of undiagnosed cases (called false negatives);
  • Reduce the number of unnecessary calls to confirm the presence of a tumor by biopsy.

And how does artificial intelligence help radiologists?

  • Helps you reduce the number of reviews you need to watch;
  • Identifies studies that require special attention and scores the studies as they are;
  • Organize according to urgency;
  • Reduces the number of undetected cancer cases;
  • Reduces the number of biopsies performed in case of doubt.

– Artificial intelligence considerably extends the possibilities of diagnostic imaging in oncology. It allows an accurate clinical decision based on standard radiological tests. We don’t use new methods, we just keep improving them. We try to find such characteristics of the image that tell us a little more than we thought so far. This means faster and more reliable diagnosis, concluded Dr. Maciej Bobowicz.

Clear up the chaos

We don’t have to be afraid of artificial intelligence. AI-based tools can extract valuable insights from large datasets and then provide actionable insights applicable in many situations.

AI technology helps make sense of the overwhelming amount of clinical data, medical literature, and information about people and work methods to facilitate informed decision-making. Using artificial intelligence, healthcare providers can learn more, interpret billions of textual and visual data in an instant to assess patient information as accurately as possible.

Source: Zdrowie.pap.pl, Katarzyna Walterska

Leave a Comment