Automating literature screening and curation with applications to computational neuroscience.

OBJECTIVE: ModelDB (https://modeldb.science) is a discovery platform for computational neuroscience, containing over 1850 published model codes with standardized metadata. These codes were mainly supplied from unsolicited model author submissions, but this approach is inherently limited. For example, we estimate we have captured only around one-third of NEURON models, the most common type of models in ModelDB. To more completely characterize the state of computational neuroscience modeling work, we aim to identify works containing results derived from computational neuroscience approaches and their standardized associated metadata (eg, cell types, research topics). MATERIALS AND METHODS: Known computational neuroscience work from ModelDB and identified neuroscience work queried from PubMed were included in our study. After pre-screening with SPECTER2 (a free document embedding method), GPT-3.5, and GPT-4 were used to identify likely computational neuroscience work and relevant metadata. RESULTS: SPECTER2, GPT-4, and GPT-3.5 demonstrated varied but high abilities in identification of computational neuroscience work. GPT-4 achieved 96.9% accuracy and GPT-3.5 improved from 54.2% to 85.5% through instruction-tuning and Chain of Thought. GPT-4 also showed high potential in identifying relevant metadata annotations. DISCUSSION: Accuracy in identification and extraction might further be improved by dealing with ambiguity of what are computational elements, including more information from papers (eg, Methods section), improving prompts, etc. CONCLUSION: Natural language processing and large language model techniques can be added to ModelDB to facilitate further model discovery, and will contribute to a more standardized and comprehensive framework for establishing domain-specific resources.

Swimming behavior and hydrodynamics of the Chinese cavefish Sinocyclocheilus rhinocerous and a possible role of its head horn structure.

The blind troglobite cavefish Sinocyclocheilus rhinocerous lives in oligotrophic, phreatic subterranean waters and possesses a unique cranial morphology including a pronounced supra-occipital horn. We used a combined approach of laboratory observations and Computational Fluid Dynamics modeling to characterize the swimming behavior and other hydrodynamic aspects, i.e., drag coefficients and lateral line sensing distance of S. rhinocerous. Motion capture and tracking based on an Artificial Neural Network, complemented by a Particle Image Velocimetry system to map out water velocity fields, were utilized to analyze the motion of a live specimen in a laboratory aquarium. Computational Fluid Dynamics simulations on flow fields and pressure fields, based on digital models of S. rhinocerous, were also performed. These simulations were compared to analogous simulations employing models of the sympatric, large-eyed troglophile cavefish S. angustiporus. Features of the cavefish swimming behavior deduced from the both live-specimen experiments and simulations included average swimming velocities and three dimensional trajectories, estimates for drag coefficients and potential lateral line sensing distances, and mapping of the flow field around the fish. As expected, typical S. rhinocerous swimming speeds were relatively slow. The lateral line sensing distance was approximately 0.25 body lengths, which may explain the observation that specimen introduced to a new environment tend to swim parallel and near to the walls. Three-dimensional simulations demonstrate that just upstream from the region under the supra-occipital horn the equipotential of the water pressure and velocity fields are nearly vertical. Results support the hypothesis that the conspicuous cranial horn of S. rhinocerous may lead to greater stimulus of the lateral line compared to fish that do not possess such morphology.