Four New Adaptive Systems for Four Medical Applications - Part 1

摘要

In this work we will try to introduce to four complex Artificial Adaptive Systems, able to be applied directly to medical field with interesting results. Each of these Systems is not a simple algorithm, but a set of adaptive systems, able each time to cooperate and to compete to reach an optimal solution. Optimal solution in medical field and in real medical data is fundamentally the capability to discover hidden features and connections that are: 1. real in the real medical world; 2. useful for medical doctors; 3. hard to discover using other and more simply techniques. The four Artificial Adaptive Systems we present in this work are: 1. An Auto-Contractive Map able to project a dataset into a weighted graph, whose connections show hidden non linear relations among data; 2. An Active Connection Matrix, named J-Net, able to show in any kind of medical image features invisible to human eyes and to the other imaging filter and techniques; 3. A new methodology to read the semantic of a set of points distributed into a 2 or 3 dimensional space, named Topological Weighted Centroid (TWC). This system should be able to approximate the origin point of epidemic and its trend, only starting from the distribution in the space of same cases (see part 2); 4. A new system, named IFAST (Implicit Function As Squashing Time), dedicated to detect the invariants from a multi-sequence of signals along the time. IFAST show to be a suitable techniques to analyse the EEG of patients to understand the quality of their brain activity (see part 2).