Tele-Monitoring for Medical Diagnosis Availability

The Built-In Medical Diagnosis Procedure

The diagnosis computational algorithm we propose consists of splitting the patient`s symptoms and signs into basic or elementary ones and grouping them with corresponding medical tests (e.g. determining the components of the diagnosis), then performing the diagnosis for each component, verifying the components functionality according to the patient`s symptoms and signs and determining the final diagnosis by analyzing the components interdependence functionality.

In order to have a better look of the medical performance, we monitor the patient`s evolution and the interaction between medical personnel and the patient in order to adjust the diagnosis and/or the medication whenever is necessary for improving the patient`s status.

All this procedure is modeled with Markov chains which require no large dimensions mainly due to the partitioning algorithm and also to the structure we propose for these Markov chains. The partitioning algorithm of patient`s symptoms has the advantage that it covers the patient`s possible illness and complications of a certain treatment by reducing non-necessary patient traffic between different hospital sections and departments for performing medical investigations and therefore, saving precious time and money for both patient and medical health care system. Like in a fault tree diagnosis system, we may say that our diagnosis model concerns partitioning the system graph (e.g. the system Markov chain model) and that allows us to avoid construction of large Markov chains for each partition and to combine the diagnosis for each partition in order to obtain the general, e.g. final, diagnosis by using a causal diagnose cognitive map (CDCM). CDCM is a Markov chain that deals both with the technical devices involved in medical diagnosis and treatment as well as with the activity of the involved medical personnel. The medical diagnosis partitioning procedure is given schematically in Figure 1. Figure 1 shows that a CDCM is built for each component as we can determine the influence of the most significant component into the final diagnosis. Basically a component diagnosis represents an interface between physician and patient and consists of several provided operations and some expected or required reactions. Our approach consists of determining the optimum diagnosis, considering accuracy and time consuming, by using a built-in test of the diagnosis components model implemented with a Markov chain.