Experts leverage cognitive data to deepen insights and apply this newfound knowledge in medical contexts through the utilization of machine learning. The outcomes will be showcased on Monday, November 13, from 2–3 p.m. Eastern Time at the Society for Neuroscience’s quarterly gathering, recognized as the primary source of cutting-edge information on brain science and health globally.
Artificial intelligence (AI) represents a sector within machine learning dedicated to empowering computers to analyze data in increasingly intricate manners. Researchers employ algorithms featuring dynamic models that can adapt and evolve independently. In scientific analysis, machine learning proves instrumental in processing extensive datasets derived from the human brain, enabling predictive modeling.
Recent studies reveal the following advancements:
- By employing neuroscience methodologies to pinpoint abnormal brain regions in medical conditions, researchers have successfully linked exercise between the thalamus and motor-related regions to symptoms of melancholy. Ayumu Yamashita from the International Advanced Telecommunications Research Institute spearheads this research.
- A study focusing on utilizing kPCA (kernel Principal Components Analysis) to forecast the progression from mild cognitive impairment (MCI) to Alzheimer’s disease (AD) showcases the capability of machine learning models in analyzing patient data. Nikita Goel from the University of Southern California leads this investigation.
- Recent software employing artificial neural networks sheds light on how cells in the visual brain interpret subtle signals from the eye, translating them into vision. Dan Butts from the University of Maryland is at the forefront of this research.
- Researchers have developed a tool combining machine learning models with traditional control theory techniques to predict deep brain stimulation (DBS) targets. This innovation, led by Maral Kasiri from the University of California, Irvine, could enhance the treatment of pediatric rigidity.
Terry Sejnowski, the Francis Crick Professor at the Salk Institute for Biological Studies and a distinguished professor at UC San Diego, who moderated the press event, emphasized the transformative impact of AI and machine learning on mental research and clinical treatments. Machine learning enables the analysis of vast brain recordings data, offering significant potential in developing innovative therapeutic approaches and treatment programs such as forecast modeling, machine-brain interfaces, and neuroimaging/neuromodulation.
Overview of the Press Event
- The seminars delve into the extensive data extracted and analyzed through machine learning. The initial studies concentrate on predictive modeling, identifying brain activity associated with conditions like depression and Alzheimer’s disease (AD), while the subsequent studies explore the intricate visual brain structure and leverage deep brain network models to advance therapies.
Unsupervised Feature Selection Method Unveils Strong Resting State Functional Connectivity Related to Major Depressive Disorder
- Researchers harnessed machine learning and neuroscience techniques to identify biomarkers of mental health. By employing three distinct feature selection methods for major depressive disorder (MDD), scientists evaluated the efficacy of these techniques in pinpointing abnormal brain regions in psychiatric disorders.
- The unsupervised-based feature selection method consistently extracted impactful connections from the test data, revealing coordinated activity between brain regions, notably the brain and motor-related areas, correlated with depressive symptoms. MRI data from 1,162 patients, including 334 with depression, facilitated this discovery.
Forecasting Transition from Mild Cognitive Impairment to Alzheimer’s Disease Using Machine Learning and 3D Head MRI
- Forecast models aim to predict the onset of mental disorders like AD through machine learning techniques. Researchers employed diverse designs to identify models capable of foreseeing the progression from mild cognitive impairment (MCI) to AD.
- By integrating statistical and biological data from over 2,448 individuals, alongside MRI scans, researchers developed a model that combines examination data with information on an AD risk gene and fundamental patient data, harnessing kPCA for enhanced predictive analysis.
Medically Constrained Deep Neural Networks Unveil Essential Visual Cortex Computations
- Convolutional deep neural networks (CNNs) represent an innovative application of artificial neural networks that elucidate how brain cells in the visual cortex process light signals from the retina.
- Through equipment learning, researchers are unraveling the transmission of visual information among brain cells in the visual cortex, shedding light on the processing of the visible universe.
Utilization of Simple Recurrent Neural Network for Deep Brain Stimulation Target Identification
- Deep brain stimulation (DBS) poses challenges in pinpointing optimal brain targets for clinicians. By leveraging machine learning to model deep brain networks in conjunction with conventional control theory and statistical techniques, based on intrinsic brain signals of specific brain regions like the basal ganglia and thalamus in children with dystonia, a tool has been developed to identify the most effective DBS targets for individuals, customizing treatment for motion disorder symptoms.
