EnsoData’s Artificial Intelligence (AI) team had the privilege to attend the 33rd Neural Information Processing Systems (NeurIPS) annual conference in Vancouver, Canada. NeurIPS is the largest AI conference in the world, attracting researchers and developers from the industry, government, and academia alike. This year’s conference — which took place from December 8 to 14 — hosted a record of over 13,000 participants, demonstrating the growing global interest in AI.
The conference covered a range of subjects, from autonomous vehicles and robots to climate change and healthcare. It included an array of activities such as poster sessions, tutorials, workshops, presentations, and an expo. Our AI team had an extraordinary opportunity to explore new developments in AI research, with a focus on healthcare. We returned to our HQ at Madison, WI with many insights regarding the state of AI in healthcare. In the following blog post, we will review our three major takeaways.
AI in healthcare is still young
With the exponentially increasing interest in AI and machine learning (ML) technologies, many have argued that the face of the healthcare industry will change drastically in the following years. Indeed, AI and ML innovations have the potential to transform the lives of patients and clinicians for the better. Despite this hype, NeurIPS 2019 has highlighted that there is still a long way to go. We couldn’t help but notice that the representation of healthcare-focused companies was relatively small. Companies such as doc.ai and anthem.ai, have stood out as two of the few companies working towards commercializing AI in healthcare. While giant multinational companies heavily research AI applications in healthcare, it appears that they are largely in the development stages of these ideas. Moving towards commercializing AI into healthcare entails great challenges, from creating a user-friendly interface, to receiving FDA clearance. Here at EnsoData, we worked hard to solve these issues and look forward to contributing more to the discussion of these challenges in the future.
Small scale data is a major challenge to AI in healthcare
Although the healthcare industry was largely under-represented in NeurIPS 2019, we were excited to see the overflow of academic research on AI and ML medical solutions. While very few addressed EnsoData’s main expertise — waveform physiological signals — many of these works focused on the field of medical imaging, offering intriguing insights. A major shortcoming of these studies, however, is the small scale in which they were done, a result of under availability of healthcare data. Utilizing big data is crucial for developing AI applications, especially as AI and ML models can be highly influenced by edge cases and outliers. To make real advancements with AI in healthcare, researchers should strive to increase the size of their datasets through collaborations with clinicians and companies. At EnsoData we test and validate our algorithms on tens of thousands of observations in order to ensure that the Waveform AI we deploy in our products is robust and unbiased. Doing so allowed us to create a trustworthy technology that clinicians can rely on.
Collaborating with medical professionals is crucial for AI in healthcare
AI in healthcare can be used to create tools that save clinicians’ valuable time and resources, allowing them to focus on the personalized care their patients deserve. To deeply impact healthcare, AI companies must collaborate closely with clinicians, gain their trust and learn how to design products to ease their workload. Encouragingly, this was the prevalent approach in NeurIPS 2019. The presentations, workshops, and hallway-conversations with the broader community of developers all highlighted the important role medical professionals play in the development and future of AI in healthcare. At EnsoData, we put an emphasis on building products that our users, medical professionals, can immediately trust and begin to use with minimum changes to their current workflow. We constantly work together with our customers to add more features to our product and make the relevant customizations.
Overall, NeurIPS 2019 was a great experience. The AI community is very passionate, collegial, and inclusive. Researchers and developers of all ages, from different parts of the world, got together to share ideas, collaborate, and guide each other. We were proud to learn that EnsoData uses state-of-the-art AI methods and has gained expertise in creating novel products that address the issues raised when commercializing AI research in healthcare. We are excited for EnsoData’s future and the future of the AI community. We look forward to working and collaborating with future partners to make sure we all create safe and fair products that benefit everyone!
