Menu Close

Neuroscience and Machine Learning: A Synergistic Approach to the Future of Medicine

If we were to simplify how the human brain functions, we could think of it as mere networks of neurons that have patterns of information flow, which enables an individual’s response to a stimulus. Essentially, we have a neural code that specifies how, for example, your muscles should contract when writing and this code is subject to change depending on the context of our surroundings. Tightening your hand’s grasp when your brain senses that you are carrying a heavy object is an example of such change. With more experiences, the stronger and more efficient your neural code becomes at responding to your surroundings. This simplification is synonymous with machine learning; the concept of using computers to act and interpret like humans do, and enhancing their ability to replicate biological processes by feeding data and information into these computers, data that is derived from observations and real-world interactions1.

Chetan Pandarinath, a biomedical engineer from the Georgia Institute of Technology used this growing interaction between Neuroscience and Machine Learning to create robotic arms for patients with paralyzed limbs, allowing them to reach out to and grab hold of objects2. How is this possible? Your brain processes data from your surroundings and in response spits out more data that gives instructions to your physiological workings. Pandarinath recorded the information neurons transmit to enable movement, and fed this data into an artificial neural network, which is essentially a computer-based replica of your brain. Doing so teaches the “artificial brain” to reproduce movement the same way your real brain would2. And it only gets better from here, the more data we can get from our actual brains and feed it into this artificial network, the better it can get at replicating our cognitive processes. In another example, machine learning is being used as a diagnostic tool to detect depression in patients. By analyzing the functional connectivity patterns of the brain of a healthy individual versus a depressed individual, and inputting these patterns to the artificial network, it allows the network to recognize “abnormal” organizational pattern that would correspond to depression3.

This is only the start of such advancements; the potential is undoubtedly present. However, there are challenges, the neural network is an extremely complex system with billions of neurons, much of which is still not yet understood. However, while the brain can help inform machine learning, machine learning can inspire more out of neuroscience research as well. For example, many processes of the brain such as learning remain unclear, and perhaps making this artificial brain replicate these processes will educate us on how the human brain can perform this very function, by filling in the missing blanks4.

This collaboration will provide us with answers that decades of research has come down to. We live in the technological age, and it is only fair to embrace machine learning and artificial intelligence in general, to help us advance our current knowledge of the brain’s functioning. We can anticipate the discovery of previously undiscovered facets of the brain, and therefore, with the age of technology comes the age of new discoveries.

References:

1Faggella, D. (2020, February 26). What is Machine Learning? Retrieved May 28, 2020, from https://emerj.com/ai-glossary-terms/what-is-machine-learning/

2Savage, N. (2019, July 24). How AI and neuroscience drive each other forwards. Retrieved May 28, 2020, from https://www.nature.com/articles/d41586-019-02212-4

3Li, X., La, R., Wang, Y., Hu, B., & Zhang, X. (2020). A Deep Learning Approach for Mild Depression Recognition Based on Functional Connectivity Using Electroencephalography. Frontiers in Neuroscience, 14. doi:10.3389/fnins.2020.00192

4Marblestone, A. H., Wayne, G., & Kording, K. P. (2016). Toward an Integration of Deep Learning and Neuroscience. Frontiers in Computational Neuroscience, 10. doi:10.3389/fncom.2016.00094