The exponential increase in the number of aircrafts and air travellers has triggered new innovations which aim to make airline services more reliable and consumer friendly. Quick and efficient maintenance actions with minimum downtime are the need of the hour. Areas that have a large potential for improvement in this regard are the real time use of diagnostic data, filtering/elimination of nuisance faults and machine learning capabilities with respect to maintenance actions. Although, numerous LRUs installed on the aircraft generate massive amounts of diagnostic data to detect any possible issue or LRU failure, it is seldom used in real time. The turnaround time for LRU maintenance can be greatly reduced if the results of the diagnostics conducted during LRU normal operation is relayed to ground stations in real-time. This enables the maintenance engineers to plan ahead and initiate maintenance actions well before the aircraft lands and becomes available for maintenance. Handling nuisance faults generated during the LRU diagnostic tests is another area with scope for improvement. The advancements in predictive analytics can be harnessed to identify the possibility of reported fault being a nuisance fault. The current method to identify nuisance faults involves a maintenance engineer performing an initiated test after the aircraft touches down. Any time spent in planning maintenance actions to rectify these faults and parts procured for the same is wasted. This paper discusses a novel method that addresses the aforementioned problems by the use of on-board automated FMEA, predictive analytics and machine learning to suggest actions for maintenance engineers. The on-board automated FMEA allows critical diagnostic data to be identified, transmitted and used in real time. Predictive analytics enables the forecasting of nuisance faults and prioritizing the reported faults. The paper also outlines the implementation challenges pertaining to data communication, security and integrity.