We report a minimally invasive, synaptic transistor-based construct to monitor in vivo neuronal activity via a longitudinal study in mice and use depolarization time from measured data to predict the onset of polyneuropathy. The synaptic transistor is a three-terminal device in which ionic coupling between pre- and post-synaptic electrodes provides a framework for sensing low-power (sub μW) and high-bandwidth (0.1–0.5 kHz) ionic currents. A validated first principles-based approach is discussed to demonstrate the significance of this sensing framework and we introduce a metric, referred to as synaptic efficiency to quantify structural and functional properties of the electrodes in sensing. The application of this framework for in vivo neuronal sensing requires a post-synaptic electrode and its reference electrode and the tissue becomes the pre-synaptic signal. The ionic coupling resembles axo-axonic junction and hence we refer to this framework as an ad hoc synaptic junction. We demonstrate that this arrangement can be applied to measure excitability of sciatic nerves due to a stimulation of the footpad in cohorts of m+/db and db/db mice for detecting loss in sensitivity and onset of polyneuropathy. The signal attributes were subsequently integrated with machine learning-based framework to identify the probability of polyneuropathy and to detect the onset of diabetic polyneuropathy.