Controlling the physiological evolution of people plays an important role in the prevention and detection of health problems. Often, the medical analyzes performed by people are stored in documents that in the long term tend to be lost, suffer wear and tear and most importantly, do not facilitate the comparison of data to assess the physiological evolution. This work proposes an intelligent scanning system for blood test documents. This type of component can be found in digital (Digitally created PDFs, scanned PDFs and images) or physical format. It is intended to extract from this type of documents only the essential information, consisting of the blood components and their concentration. Optical Character Recognition (OCR) Machine Learning (ML) Kit was used to extract relevant information from documents in the case of scanned PDF's (SPDFs), images, and physical documents, and Apache's PdfBox for digitally created PDF's (DCPDFs). To filter the data and associate the blood constituents detected with the respective concentrations, a condition tree was developed. In the end, the methods used were able to detect an average of 95.38% of the blood compounds present in the different document formats. On average, 87.63% of the concentrations were correctly associated with the detected compounds.