Waseem Abbas

- To create and validate diagnostic tools powered by machine learning for better diagnosis of auto-immune diseases, models trained on omics and clinical data - Drug targets discovery: To discover novel drug targets powered by machine learning and multi-modal data

- In-silico modeling of striatal development in Huntington patients - Based on the in-silico disease models, discover genes that can be potential targets for gene therapies. (Managed to shortlist 400+ potential genes) - Meaningfully integrate gene datasets from sequencing, open arrays and multi-electrode arrays, calcium imaging - Explainable models for predicting COVID-19 severity from CTs, CXRs, Ultrasounds, MRIs and EMR - Designing domain adaptation strategies for egocentric vision

- Spatio-Temporal U-Net: Designed a new architecture of U-Net by using convolutional LSTM layers which enabled the U-Net to learn temporal and spatial features achieved state of the art results on Neurofinder Challenge - Spatio-temporal limbs detection and tracking for behaviour analysis: Developed a two-staged limbs detection and gesture tracking pipeline based on convolutional neural networks (CNNs), long short term memory networks (LSTMs), and optical flow

- Designed a pattern recognition based demodulation scheme for ultrasonic communication which extended communication range from tens of meters to hundreds of meters - Designed and developed an educational and interactive GUI for machine learning beginners https://bit.ly/2YEtQwk - Acoustic fingerprinting: An audio signal processing and neural network based pipeline to detect a specific acoustic event in ambient noise - Assisted image processing and pattern recognition courses

- Developed a pipeline to predict the return on insurance claims - Helped the company to identify insurance claims which had high chance of success