Numerous analytical techniques for life sciences have been at the basis of the improvement of medical diagnostics and treatments. Current challenges in bioanalytics are related to the conception of methods enabling large amount of information to be extracted from biological samples, in order to be translated into concrete solutions for the patient. From this point of view, microfluidics constitutes a key element to automate, accelerate and standardize bioanalytical tests. At the same time, obtaining precise information from the analyte of interest is crucial to specifically address diagnoses and therapies. Since the high flexibility of optical readouts can give access to a multitude of diverse aspects of biological matter in multiplexed fashion, in this thesis we have chosen to integrate microfluidics and multimodal optics to address specific problems in translational bioanalytics: drug screening and tumor diagnostics. In the first part of the work, we approached the problem of drug screening on living organisms, and especially the possibility to provide as much information as possible on potential mechanisms of action and clinical targets. For this, we aimed at developing a method to study the model organism Caenorhabditis elegans, i.e. a small soil worm that is widely used by researchers to study physiopathology at whole-organism scale. In fact, while offering the systemic information that is missing with cellular models, this small organism is much easier to handle and grow in large amounts compared to rodents, for which the costs and the ethical concerns are substantial. Moreover, its large genetic homology with humans supports translational analysis and its genetic manipulation is well mastered to model human diseases. Thus, we developed a microfluidic array to select small separated populations of C. elegans specimens, and combined fluorescence and bright-field imaging along with high-throughput feature recognition and signal detection to identify the mode-of-action of an antibiotic. Also, we demonstrated real-time, very large field-of-view capability on multiplexed motility assays for the assessment of the dose-response relation of an anesthetic. Furthermore, to enable on-site antibiotic testing, we explored a method to empower low-magnification optical systems with dielectric micro- and macro-spheres for the detection of pathogens in liquid media. Such dielectric objects offer potential solutions to increase the resolution and the sensitivity of imaging systems in a cost-effective fashion. We first characterized the parameters that govern the resolution increase in the case of microsphere-assisted incoherent microscopy and demonstrated it by imaging silicon microstructures and fluorescent bacteria. Then, we developed a method to easily fabricate microfluidic chips that integrate custom designs and precise location of highly-diffractive macro-spheres, and used it for on-chip detection of bacteria in a flow. In the second part of the work, we aimed at developing a technique to assess the status of individual tumor and immune cells in the conventional tissue sections used in the clinics. The first step was to combine fluorescence-based image segmentation with automated microfluidics to quantify, at the single-cell level, the overexpression of two proteins used as biomarkers for breast cancer diagnostics. We first assessed our method on biological controls such as breast cancer cell lines, and then combined it with ...