What are your variables: dependent (x) and independent (y). Your y is dementia; to conclude that are is dementia, you need to define y1, y2, ...yi. the factors contributed to dementia would be x1, x2, ..., xi. Once you establish relationship between x and y then you can construct predictive model: y = a + b1x1 + b2x2 + ... If you speculate dementia is linear or if non-linear explore other types of distribution that may be best fit for mental degradation.