Abstract: A computing system for determining a systematic training strategy for the user is provided. The computing system includes a user device that uses one or more sensors to obtain partial pressure of oxygen (PO2) levels of a user over a period of time. The user device provides previous PO2 levels to a personalized erythropoiesis model generation computing platform. The computing platform obtains individualized user data for the user indicating or more previous hematocrit and/or hemoglobin measurements for the user. The computing platform determines an individualized erythropoiesis model for the user based on the one or more previous hematocrit and/or hemoglobin measurements and the previous PO2 information, and employs the individualized erythropoiesis model to determine predicted hematocrit and/or hemoglobin measurements. The computing platform performs one or more actions based on the one or more predicted hematocrit and/or hemoglobin measurements.