Post by Leigh Christopher
What did we learn?
2021 saw a big step forward in the use of Machine Learning in neuroscience. It’s no secret that the brain’s complexity is vast, and although scientists have come closer to understanding how it functions, there is still a long way to go. To truly understand some of the mechanisms of the brain, for example how a disease progresses or how we make complex decisions, machine learning can be a valuable tool. Research this year showed that neural networks - otherwise known as ‘deep learning’ - could be used to decode meaningful information from raw neural recordings - such as an animals’ spatial location, speed, or direction, highlighting how powerful machine learning can be in connecting complex neural activity to specific behaviors. Another study used neural networks to better understand how different brain areas process visual information. They were able to predict exactly how the brain would respond to particular stimuli and confirm their hypothesis that the brain responds to specific categories of visual information. Other research this year focused on how to apply machine learning to advance personalized medicine. One study in particular applied machine learning techniques to predict individual drug responses and outcomes in temporal lobe epilepsy, taking into account individual disease characteristics. Rather than classify patients into specific groups, they were able to use the variability in their data to provide more nuanced insights into how patients might respond to various treatments - an important step towards personalized medicine that could apply to a wide range of diseases.
What's next?
2021 was an important year for progressing our understanding of the brain, and further incorporating machine learning techniques into research methodologies. Although there was a big step forward, we are only at the tip of the iceberg in terms of the potential for machine learning to change the way we conduct neuroscience research, and develop real-world applications to advance science and medicine. As we saw this year, machine learning can be used to link complex brain activity to a specific behavior, to help us understand how the brain operates at a system-wide level, to better characterize diseases, and advance personalized medicine. The applications of machine learning are broad, however, there is a need to better translate the insights from these powerful techniques into impact. 2022 will hopefully be a year in improving the interpretability of machine learning for widespread use amongst the scientific community.
