Droplet based microfluidics integrated with machine learning

Droplet based microfluidics (DBMF) has gained huge recognition in the recent years for performing micro-reactions in droplets with high throughput, sensitivity, specificity and minimum cross-contaminations. This technology enables the researchers to realize highly reliable and rapid detection and screening applications in various fields. The high-throughput nature of droplet microfluidics generates large amounts of valuable but complex droplet dataset. Deeper analysis of this intricate droplet data is very essential for detection, classification, characterization and quantification of reactions/content inside the droplets. This can be carried out by Machine Learning (ML), which has proven itself in processing and providing deeper insights and precise predictions of relatively large amounts of complex data with shorter analysis times and exceptional accuracy. The analytical tools of ML enable to imbibe automation and control of many such diagnostic platforms, including DBMF, with minimum human intervention. In recent times, the potential of ML has been explored in microfluidic technology as well to tackle challenges in biomedical and biotechnological applications. The synergy of both the fields, DBMF and ML, helps in development of optimized and automated tools with higher accuracy for numerous applications. Specifically, this enables complete comprehension of the field to eventually realize a truly microfluidic total analysis system (µTAS). This work comprehends a general review emphasizing the implementation of different ML models with DBMF to automate various activities such as fluid control, droplet size prediction, recognition of flow pattern and identification, classification and sorting of droplets in a microfluidic device.