Endocrine Abstracts

Conventionally, physicians adjust LT4 dose by clinical estimation based on their experiences and available guidelines; however, no computational algorithm exists to support their decisions. The objective of this study is to build an artificial decision support system utilizing fuzzy logic to propose a proper LT4 dosage regimen. We used THYROSIM software as the model of human thyroid hormone regulation to simulate a virtual thyroidectomized patient, setting the secretion rates of thyroid gland at 1%. A T-S fuzzy control algorithm was developed in MATLAB software to track TSH and T3 concentrations to their objective setpoints by adjusting LT4 dosage scheme. The result of our proposed LT4 dosage for the virtual patient was then compared with a conventional guideline implementing into THYROSIM software. The dosage of 165 µg LT4/day for our virtual patient results to 14 days lag time to achieve the objective setpoints of TSH and T3 while the proposed dosage by T-S fuzzy control algorithm could reach our objective in 8 days. Moreover, in contrast to fuzzy control system TSH concentraion in convetional therapy exceeded the upper bound limit for a priod of 7 days. Fuzzy logic could assist physicians by proposing a more proper LT4 dose scheme with an accuracy exceeding that of an expert provider. Therefore, a more proper treatment could be delivered by the clinicians using the decision tree of fuzzy logics.