This study is an exploratory analysis of the timely detection of stress and anxiety in university students, which seeks to lay the foundations for an early stress and anxiety alert scheme focused on characterizing multisignal patterns and relationships to guide the design of decision rules and thresholds for future detection, through the integration of physiological signals captured by a wearable (Fitbit Inspire 3) and daily self-reports in m-Path. A total of 32 students from public and private institutions participated during a school term. Physiological variables were collected such as heart rate variability (RMSSD (Root Mean Square of Successive Differences), NREMHR (Non-REM Heart Rate), entropy; LF/HF bands), sleep (minutes in deep sleep, resting heart rate, nighttime restlessness, “sleep score”), SpO2 (average nocturnal peripheral oxygen saturation and per minute), heart rate (number of heartbeats per minute), and respiratory rate. After preprocessing and daily aggregation, similarity matrices between students were constructed for each signal (cosine distance), which were normalized and averaged to obtain a global correlation similarity across variables, visualized with heatmaps and used for clustering. When contrasting said similarity with the absolute differences in reported stress and anxiety, a positive association of small magnitude was observed: in practical terms, the greater the physiological similarity between pairs of students, the closer their self-perceptions of stress and anxiety tend to be, although with considerable interindividual variability. The work will continue with an intra-subject (per-student) focus, instead of contrasting the entire study group, using sliding windows and multi-signal rules to detect personal patterns of stress and anxiety and support early alerts in academic contexts.