Pranav Rajpurkar is a researcher in medical artificial intelligence and a professor of biomedical informatics at Harvard Medical School. He came up through Stanford’s AI group, where he worked on both natural language processing and clinical imaging.
He is the first author of the 2017 CheXNet paper, which trained a 121-layer convolutional network on more than 100,000 chest X-rays and reported pneumonia detection that exceeded the average radiologist on the test set. The work, with Andrew Ng and others, became one of the most widely cited demonstrations of deep learning applied to medical imaging and helped popularize the use of large labeled radiology datasets.
Earlier, as a Stanford student, Rajpurkar co-created SQuAD, the Stanford Question Answering Dataset, which became a standard benchmark for reading-comprehension systems and is documented separately in this library. His later research has focused on building and rigorously evaluating clinical AI across imaging and other modalities.
For a general reader, Rajpurkar is a useful figure to know because his career spans two of the threads in this library: the benchmark datasets that drove NLP progress and the medical-imaging models that brought deep learning into healthcare.