Therapeutic Drug Monitoring (TDM) is considered one of the most successful approaches to personalized medicine. However, the accuracy requirements on the quantification of drug concentration and the challenges in the dose adjustment limit the application of TDM only to specialized facilities. In order to spread TDM at the family doctor cabinet it is required to perform the concentration measurements by means of point-of-care tests. A possible solution to this analytical challenge is the employment of molecular sensing systems in the form of top bench or hand held devices with disposable cartridges. The implementation of an analytical-grade biosensor requires the determination of the sensing molecule, the development of a reliable analytical assay and the realization of the physical support hosting the sensing surface. In the frame of this thesis, aptamers have been identified as promising sensing molecules for their high stability and resistance, low molecular weight and reproducible synthesis. A protocol for selection of aptamers against efavirenz has been implemented and tested. Since Tween 20 was added to the selection buffer to increase efavirenz solubility, a preliminary negative selection is performed on each cycle to remove sequences that display affinity for the solubilizing agent. After nine selection cycles no enrichment of the pool was detected, nonetheless the results gave useful indications for a future implementation of the protocol with an improved negative selection. As far as analytical assay development is concerned, in this thesis the testing of an aptamer-based assay on patient samples is presented for the first time. The assay is label-free and employs Surface Plasmon Resonance as detection technique. The limit of detection is 0.15 µM with a variability of 7% and a linear response up to 10 µM. The error percentage on accuracy on patient samples is 50% (30% above 1 µM). This is referable to two reasons: (i) the non-specific interactions of the matrix with the sensor surface dominate at lower concentration, (ii) the calibration is performed using a different serum, which leads to matrix effects lowering the assay performance. While these results are not compatible with clinical applications (15% tolerance), however they provide further indications for assay improvement. For instance, a standard addition calibration approach have been attempted which resulted in the suppression of the matrix effect. The development of a reliable physical surface for biosensors has been investigated in the frame of an electrochemical detection technique. While the fabrication of electrodes in the micro-scale is well established, the challenge resides in the robustness of their passivation, which should withstand chemical modification and prolonged exposure to liquid environment. Five different passivation layers have been tested: sputtered SiO2,low-pressure chemical vapour deposition (LPCVD) low-temperature oxide (LTO), Parylene C, SU-8, and dry-films. Parylene C and SU-8 have shown superior performance both in hydrolysis tests and impedance tests. Parylene C passivated electrodes have been then employed for electrochemical-based measurements with ferrocene-labeled alkanethiols, as preliminary tests for biosensing applications.