Example data for a multi-source heterogeneous data-driven multi-granularity sequential three-way decision method for medical diagnosis
Description
This dataset contains the illustrative example data of the rheumatoid arthritis (RA) diagnosis case study used in the paper "A multi-source heterogeneous data-driven multi-granularity sequential three-way decision method for medical diagnosis". The data describe a three-stage sequential diagnostic process involving 15 patients (x1-x15) and a healthy reference profile (x*): (1) Stage 1 - Physical examination: heterogeneous evaluation information of five attributes (pain severity and swelling severity expressed by double hierarchy linguistic terms, number of painful joints, symmetry of joint involvement, and morning stiffness duration), the corresponding normalized distance measures, conditional probabilities, relative loss-based thresholds, and the first-stage three-way decision results (Tables 1, 2, 4); (2) Stage 2 - Blood test: observed values of four laboratory indicators (rheumatoid factor, anti-cyclic citrullinated peptide antibody, C-reactive protein, and erythrocyte sedimentation rate) for the patients carried over from the first-stage boundary region, together with the normalized distances, conditional probabilities, thresholds, and decision results (Tables 5-7); (3) Stage 3 - Medical imaging: 32-dimensional ResNet50 feature vectors extracted from joint color Doppler ultrasound images, cosine distances, conditional probabilities, thresholds, and decision results (Tables 8-9); (4) Model parameters: monthly patient numbers and diagnosis rates at each stage, hesitation aversion coefficients, stage weights, attribute definitions with data types, healthy reference values and weights, and the double hierarchy linguistic term sets (Table 3 and related settings); All data are provided in a single Excel workbook, fully static and without formulas. Values are transcribed as reported in the paper.
Files
Institutions
- Jiangxi University of Finance and EconomicsJiangxi, Nanchang