Digital Insurance Adoption and ANN-Based Purchase Prediction: A Consumer-Rights Protection Perspective

Digital insurance expands access but raises concerns about information, choice, privacy, and fair treatment. Using 15,000 secondary records—10,000 insurer records and 5,000 public survey/social-media records—this study predicts a binary online-insurance purchase outcome from age, gender, income, education, and occupational risk. A multilayer perceptron ANN achieves 87.65% accuracy, 85.20% recall, 86.41% F1, and 0.92 AUC. Against random forest, the closest supervised benchmark, accuracy is 5.55% points higher and AUC 0.04 higher. Income is the strongest displayed association; consumer rights are interpretive rather than measured constructs, supporting transparent, privacy-conscious, non-exclusionary use.