Donating blood is a selfless gesture with the possibility of saving lives. Several techniques exist for giving blood. Each kind contributes to meeting certain medical needs. Donating blood is completely safe. By providing clean, single-use equipment to each donor, we can ensure that no one gets an infection from donating blood. Most healthy people can give half a liter of blood (a pint) without risking their health. The body quickly replaces the fluids lost after donating blood. In 2 weeks, the body will replenish the lost red blood cells. For the safety of donors and patients, it is essential to identify and manage risks in the donation process and blood products. In this study, we create models for clumped datasets by enhancing performance significantly. The need for transfusions as a result of accidents, procedures, diseases, and other circumstances is the driving force behind this work effort. Realistic assessments of the number of blood donors allow healthcare professionals to establish strategies for motivating people to donate blood to meet needs. With the goal of predicting high-risk donors, this research attempts to assess the risk analysis of blood donors by using DL (deep learning) techniques based on a recurrent neural network (RNN) and a co-relational network to automatically identify blood donors. The model incorporates RNNs to track patients’ progress over time, improving the accuracy of future projections. In order to provide more precise risk assessments and appropriate donor referrals, a probabilistic neural network (PNN) is fed data from a co-relational neural network (CRNN) that has identified the most crucial aspects of blood donation. The experimental findings demonstrate improved accuracy in assessing total risk in both donors and recipients.