This study addresses the challenge of multi-dimensional and small gas sensor data classification using a gelatin–carbon black (CB-GE) composite film sensor, achieving 91.7% accuracy in differentiating gas types (ethanol, acetone, and air). Key techniques include Principal Component Analysis (PCA) for dimensionality reduction, the Synthetic Minority Over-sampling Technique (SMOTE) for data augmentation, and the Support Vector Machine (SVM) and K-Nearest Neighbor (KNN) algorithms for classification. PCA improved KNN and SVM classification, boosting the Area Under the Curve (AUC) scores by 15.7% and 25.2%, respectively. SMOTE increased KNN’s accuracy by 2.1%, preserving data structure better than polynomial fitting. The results demonstrate a scalable approach to enhancing classification accuracy under data constraints. This approach shows promise for expanding gas sensor applicability in fields where data limitations previously restricted reliability and effectiveness.

Thermal sensors are widely used in medical, industrial and other fields, where the requirements for high sensitivity and portability continues to increase. Here we propose a suspended bridge structure fabricated using MEMS, which effectively shrinks the size and reduces heat loss. This study reviews current sensor-related theories of heat conduction, convective heat transfer and thermal radiation. Heat loss models for suspended and non-suspended bridge structures are established, and finite element analysis is conducted to evaluate their thermal performance. The thermal performance of the suspended bridge structure is further validated through infrared temperature measurements on the manufactured sensor device. Theoretical calculations demonstrate that the proposed suspension bridge structure reduces heat loss by 88.64% compared with traditional designs. Benefiting from this improved heat retention, which was also confirmed by infrared thermography, the thermal sensor fabricated based on the suspension bridge structure achieves an ultra-high sensitivity of 0.38 V/W and a fast response time of less than 200 ms, indicating a high accuracy in thermal characterization. The correlation coefficient obtained for the sensor output voltage and input power of the sensor is approximately 1.0. Based on this design, multiple microfluidic channels with suspended bridge structures can be integrated to realize multi-component detection, which is important for the development of multifunctional biomedical detection.