The expectation is to develop bridges with service life of 100 years or more. Prefabricated bridge elements and systems (PBES) are widely used to accelerate bridge construction because prefabrication is expected to offer durable components or systems to build bridges with such a long service life. However, field cast connections between PBES are the weakest link in terms of durability. Hence, to develop durable bridges not only the use of durable PBES but also the use of connection details and materials with a record of durable performance and implementation of the best practices in construction, quality assurance/quality control (QA/QC), and maintenance are needed. When a bridge with PBES is considered as a structural system, deterioration starts at the field cast connections. Hence, it is vital to identify the onset of deterioration at connections before visual signs are developed to initiate maintenance actions to be effective and efficient in arresting further deterioration. When a bridge is considered, the deck is the shelter of the structure which is subjected to severe loads due to exposure and traffic. Hence, detecting the onset of deck connection deterioration is vital to ensure a long service life of a bridge. This paper presents an Artificial Neural Network (ANN) based approach for detecting onset of precast component connection deterioration. As a prototype, a full-depth deck panel systems is used. The stress data recorded at a transverse connection is used. The results are promising; however, need further investigations to establish the distress thresholds.