Towards Automatic Diagnosis of Diseases from Biological Signals

Abstract

It can be argued that there is no greater goal in science than the one that leads to the betterment of human health-care and health in general. At the beginning of our lives, most of us has been given a clean bill of health that we often do our best to keep. It is reasonably safe to assume that, sooner or later, the environment catches up on us and we suffer an injury, illness or other health-impacting conditions. Many of those are treatable and curable without too much of an effort, many are still a great challenge and cannot be dealt with.

It is the goal of medical sciences to find ways of curing any condition that the human body can get into. Fortunately, medicine has found a great ally in the fight - computer science. Combining technology with biology provides previously unknown methods of diagnosing, studying and possibly repairing any damage to the body, no matter its cause. However, given the complexity of biological systems in general, much less the human body, many problems are still to be solved, presenting a challenge that in some (and not rare) cases cannot be overcome with today's knowledge. It falls on the scientists to find ways of converting challenging to trivial, or at least manageable.

This dissertation thesis provides an insight to the current state of the biological signal processing research area, focusing on the brain, the heart and the physical manifestations (movement) of the human body. Quite often, these research areas are treated differently and separately, with much less research focused on actually combining them together to provide a more complex, yet more capable system. For this reason, the thesis suggests an approach to use some or all of the above mentioned biosignals in a single illness-diagnosing tool.