How it works
A series of keywords are extracted from the anamnesis text.
A series of 630 metabolomic quantities are extracted from blood analysis
A Convolutional Neural Network (CNN) with inputs the key-words and the selected principal components is used as a classifier of patients with risk of PD. We identify possible patterns in the data of subclinical patients in common with early PD subjects.
If NO Risk for PD is detected, the medical doctor can take as granted the result for further diagnosis
If a risk for PD is detected, [18F]-DOPA PET scan is recommended. We extract textural features from the PET image and We identify possible patterns in the data of subclinical patients in common with early PD subjects.
The final result is a risk assessment for PD for a patient in subclinical condition, showing only generic non-specific symptoms (keywords).