People have paid more and more concern to their health issues, however, a rapid diagnosis is not always available 24/7 hrs for everyone. What’s more, the very definition of many symptoms are still in dispute; for example any physiology index is not absolutely unique criteria for the final judging when the values are vibrating near certain defined levels or just closing to them. Sub-health status is very common for modern men, thus how to quickly detect it and find some remedies is a very pressing questions, since the early interference may stop it developing into real diseases.

Human-body is highly nonlinear system, we are only at the fundamental stage in fully understand it. Neural networks and fuzzy inference system are quite good in dealing with high dimensional nonlinear systems from modeling point of view, and their combination ANFIS is even better in certain aspects. Thus it is used in Medibot to predict the healthy status and give reference remedies through learning/ training.

People in the far east have developed a method called Traditional Chinese Medical (TCM)- diagnosing symptoms through gazing, Q/A and pulsating measurement. It shows that some groups of its methods do prove effective especially when coping with Sub-healthy status. With an analogy to system identification method from automation, the diagnose and treatment processes are very similar to identify a process with just some of the outputs(patterns with profound human-biological index/phases, such as pulse; skin, tongue shape and color, torso shape ; Q/A by both patient and doctors etc.) by human , and to “stabilize ” the normal body status by giving formulae.

So, the whole picture is like this: Obtaining 3.65M (million) samples/each year for 3 to 5 years, for training /learning; and then using the model to test and gauge for another 2 to 3 years for 3.65M samples/ each year before it could finally step into the market at least to the far-east personal health consumer market. The samples could be available from China’s TCM hospitals(e.g. 100 hospitals each with 100 patients diagnose samples).

As it is shown in the pictures that the Medibot has a smartfinger to sense the pulse status, and high resolution digital camera to snapshot the tongue’s shape and color etc. then transmit to its main calculation FPGA-circuits board which is embedded with trained ANFIS, the system will then automatically output the diagnosed result including the symptoms and remedies etc.

At the training stage of this branch of inputs, we need to construct a Wavelet package to fully detest its modes from different resolution levels. The tongue’s shape and color is perhaps the easiest one to be understand since image processing is in rather mature stage. Finally we should reemphasize that the effectiveness of many of TCM formulae has been gradually proved and there are a lot of printed English version works as one could do google them. Also, the samples from other ethnic group are Quite important.