Sectors such as logistics or supply chain have already benefitted greatly from digital processes and their efficiency. Forecasts show that applications such as artificial intelligence (AI) will also have a very strong impact on the healthcare industry. For example, machine learning can be used to support the evaluation of X-ray images and thus make more precise diagnoses. AI is also improving the early detection of Alzheimer´s disease (with an accuracy of 82 to 90 percent) and breast cancer. In the second application of cancer, AI can not only predict the disease at an early stage, but also suggest a suitable therapy, for example how a patient will react to chemotherapy.
A project with good prospects
The use of artificial intelligence will soon be indispensable in health research itself – especially when it comes to evaluating a large number of data sets in order to determine possible correlations or causalities. A current project shows how such modern technologies can support research in a targeted manner. The project focuses on rare neuromuscular diseases that are a subgroup of neurological diseases and often have a genetic cause. These diseases can have a fatal course. Early diagnosis is therefore important in order to take possible measures at an early stage.
Within the framework of the German program “NRW Lead Market Competition Life Sciences,” coordinated by the Ministry of Science together with the ERDF.NRW (European Regional Development Fund), gene and protein signatures are to be investigated as a diagnostic system for neuromuscular diseases. The pharmaceutical industry is not very interested in this research because it is estimated that in Germany “only” up to 2.6 out of 10,000 children are affected by these diseases. There are no official statistics and therefore not much lucrative business potential from the perspective of the pharmaceutical companies.
The background of the disease: Neuromuscular diseases are caused by mutations in several hundred genes, among other things. Therefore, it makes sense to analyze disease-relevant genes. Currently, however, despite intensive gene analyses, about 50% of neuromuscular cases remain unexplained. Interestingly, the combination of gene and protein analysis in patients with neuromuscular diseases has led to the identification of first genetic variants. Therefore, the analysis of protein compositions using patient material has increasingly entered research. So far, however, there is hardly any clinical data that can be combined and used as a basis for developing algorithms for targeted pattern recognition. Therefore, the disease can currently only be diagnosed by gene analysis or muscle biopsy – two costly procedures that are usually applied late in the course of the disease. A much cheaper variant would be so-called proteomic analyses, which examine proteins by taking a blood sample and can be used very early and inexpensively.
In order to use this diagnostic variant, however, a connection must first be established between a too high or too low protein value and a specific gene mutation, which in turn suggests a neuromuscular disease.
Algorithms and their prospects
This is where artificial intelligence comes in. To establish a relationship between genes and proteins and at the same time consider additional patient data, modern data analytics methods are required. The prerequisite for this is the examination of various statistical methods, as well as the correct data preparation. Using a specially generated algorithm, which is “trained” using the data sets of many collaborating clinical partners, it becomes possible to determine patterns of gene and protein correlation that predict the genetic defect as accurately as possible. This very reliable diagnosis using modern data analytics will significantly improve the consultation and timeliness of treatment.
The project will be carried out over a period of three years and is further proof of the importance that technology, especially AI, can have in healthcare and related research. Analyzing and linking millions of data in rapid time can bring new insights and completely eliminate the manual effort – to eliminate diseases worldwide.
What applications of artificial intelligence are you familiar with from the healthcare sector? Where do you see opportunities and risks?