Predicting blood-brain barrier permeability of molecules with a large language model and machine learning.

Predicting the blood-brain barrier (BBB) permeability of small-molecule compounds using a novel artificial intelligence platform is necessary for drug discovery. Machine learning and a large language model on artificial intelligence (AI) tools improve the accuracy and shorten the time for new drug development. The primary goal of this research is to develop artificial intelligence (AI) computing models and novel deep learning architectures capable of predicting whether molecules can permeate the human blood-brain barrier (BBB). The in silico (computational) and in vitro (experimental) results were validated by the Natural Products Research Laboratories (NPRL) at China Medical University Hospital (CMUH). The transformer-based MegaMolBART was used as the simplified molecular input line entry system (SMILES) encoder with an XGBoost classifier as an in silico method to check if a molecule could cross through the BBB. We used Morgan or Circular fingerprints to apply the Morgan algorithm to a set of atomic invariants as a baseline encoder also with an XGBoost classifier to compare the results. BBB permeability was assessed in vitro using three-dimensional (3D) human BBB spheroids (human brain microvascular endothelial cells, brain vascular pericytes, and astrocytes). Using multiple BBB databases, the results of the final in silico transformer and XGBoost model achieved an area under the receiver operating characteristic curve of 0.88 on the held-out test dataset. Temozolomide (TMZ) and 21 randomly selected BBB permeable compounds (Pred scores = 1, indicating BBB-permeable) from the NPRL penetrated human BBB spheroid cells. No evidence suggests that ferulic acid or five BBB-impermeable compounds (Pred scores < 1.29423E-05, which designate compounds that pass through the human BBB) can pass through the spheroid cells of the BBB. Our validation of in vitro experiments indicated that the in silico prediction of small-molecule permeation in the BBB model is accurate. Transformer-based models like MegaMolBART, leveraging the SMILES representations of molecules, show great promise for applications in new drug discovery. These models have the potential to accelerate the development of novel targeted treatments for disorders of the central nervous system.

Modeling Coil-Globule-Helix Transition in Polymers by Self-Interacting Random Walks.

Random walks (RWs) have been important in statistical physics and can describe the statistical properties of various processes in physical, chemical, and biological systems. In this study, we have proposed a self-interacting random walk model in a continuous three-dimensional space, where the walker and its previous visits interact according to a realistic Lennard-Jones (LJ) potential uLJr=εr0/r12-2r0/r6. It is revealed that the model shows a novel globule-to-helix transition in addition to the well-known coil-to-globule collapse in its trajectory when the temperature decreases. The dependence of the structural transitions on the equilibrium distance r0 of the LJ potential and the temperature T were extensively investigated. The system showed many different structural properties, including globule-coil, helix-globule-coil, and line-coil transitions depending on the equilibrium distance r0 when the temperature T increases from low to high. We also obtained a correlation form of kBTc = λε for the relationship between the transition temperature Tc and the well depth ε, which is consistent with our numerical simulations. The implications of the random walk model on protein folding are also discussed. The present model provides a new way towards understanding the mechanism of helix formation in polymers like proteins.

HDOCK update for modeling protein-RNA/DNA complex structures.

Protein-nucleic acid interactions are involved in various cellular processes. Therefore, determining the structures of protein-nucleic acid complexes can provide insights into the mechanisms of the interactions and thus guide the rational drug design to modulate these interactions. Due to the high cost and technical difficulties of solving complex structures experimentally, computational modeling such as molecular docking has been playing an important role in the study of molecular interactions. In order to make it easier for researchers to obtain biomolecular complex structures through molecular docking, we developed the HDOCK server for protein-protein and protein-RNA/DNA docking (accessed at http://hdock.phys.hust.edu.cn/). Since its first release in 2017, HDOCK has been widely used in the scientific community. As nucleic acids may include single-stranded (ss) RNA/DNA and double-stranded (ds) RNA/DNA, we now present an updated version of HDOCK, which offers new options for structural modeling of ssRNA, ssDNA, dsRNA, and dsDNA. We hope this update will better help the scientific community solve important biological problems, thereby advancing the field. In this article, we describe the general protocol of HDOCK with emphasis on the new functions on RNA/DNA modeling. Several application examples are also given to illustrate the usage of the new functions.

Federated learning for predicting clinical outcomes in patients with COVID-19.

Federated learning (FL) is a method used for training artificial intelligence models with data from multiple sources while maintaining data anonymity, thus removing many barriers to data sharing. Here we used data from 20 institutes across the globe to train a FL model, called EXAM (electronic medical record (EMR) chest X-ray AI model), that predicts the future oxygen requirements of symptomatic patients with COVID-19 using inputs of vital signs, laboratory data and chest X-rays. EXAM achieved an average area under the curve (AUC) >0.92 for predicting outcomes at 24 and 72 h from the time of initial presentation to the emergency room, and it provided 16% improvement in average AUC measured across all participating sites and an average increase in generalizability of 38% when compared with models trained at a single site using that site's data. For prediction of mechanical ventilation treatment or death at 24 h at the largest independent test site, EXAM achieved a sensitivity of 0.950 and specificity of 0.882. In this study, FL facilitated rapid data science collaboration without data exchange and generated a model that generalized across heterogeneous, unharmonized datasets for prediction of clinical outcomes in patients with COVID-19, setting the stage for the broader use of FL in healthcare.

Federated Learning used for predicting outcomes in SARS-COV-2 patients.

'Federated Learning' (FL) is a method to train Artificial Intelligence (AI) models with data from multiple sources while maintaining anonymity of the data thus removing many barriers to data sharing. During the SARS-COV-2 pandemic, 20 institutes collaborated on a healthcare FL study to predict future oxygen requirements of infected patients using inputs of vital signs, laboratory data, and chest x-rays, constituting the "EXAM" (EMR CXR AI Model) model. EXAM achieved an average Area Under the Curve (AUC) of over 0.92, an average improvement of 16%, and a 38% increase in generalisability over local models. The FL paradigm was successfully applied to facilitate a rapid data science collaboration without data exchange, resulting in a model that generalised across heterogeneous, unharmonized datasets. This provided the broader healthcare community with a validated model to respond to COVID-19 challenges, as well as set the stage for broader use of FL in healthcare.

Report of clinical bone age assessment using deep learning for an Asian population in Taiwan.

INTRODUCTION: A deep learning-based automatic bone age identification system (ABAIs) was introduced in medical imaging. This ABAIs enhanced accurate, consistent, and timely clinical diagnostics and enlightened research fields of deep learning and artificial intelligence (AI) in medical imaging. AIM: The goal of this study was to use the Deep Neural Network (DNN) model to assess bone age in months based on a database of pediatric left-hand radiographs. METHODS: The Inception Resnet V2 model with a Global Average Pooling layer to connect to a single fully connected layer with one neuron using the Rectified Linear Unit (ReLU) activation function consisted of the DNN model for bone age assessment (BAA) in this study. The medical data in each case contained posterior view of X-ray image of left hand, information of age, gender and weight, and clinical skeletal bone assessment. RESULTS: A database consisting of 8,061 hand radiographs with their gender and age (0-18 years) as the reference standard was used. The DNN model's accuracies on the testing set were 77.4%, 95.3%, 99.1% and 99.7% within 0.5, 1, 1.5 and 2 years of the ground truth respectively. The MAE for the study subjects was 0.33 and 0.25 year for male and female models, respectively. CONCLUSION: In this study, Inception Resnet V2 model was used for automatic interpretation of bone age. The convolutional neural network based on feature extraction has good performance in the bone age regression model, and further improves the accuracy and efficiency of image-based bone age evaluation. This system helps to greatly reduce the burden on clinical personnel.

Accuracy of A Handheld Accelerometer-Based Navigation System for Femoral and Tibial Resection in Total Knee Arthroplasty.

Restoration of mechanical axis in total knee arthroplasty (TKA) is correlated with improved implant survivorship. We assessed the accuracy and required surgical time using a hand-held accelerometer-based navigation system for TKA. Data collected on 53 patients included assembly, resection, and tourniquet times. Implant alignment and mechanical axis were measured on radiographs. Femoral alignment was 0.29° ± 2.2° varus. Tibial alignment was 0.09° ± 1.4° valgus. Postoperative mechanical axis was 0.2° ± 2.1° varus. Malalignment rates for the femur, tibia, and axis were 13%, 3.8%, and 17%, respectively. Average time for pinning and navigating was 3.6 minutes for the femur and 2.6 minutes for the tibia; mean tourniquet time was 62 minutes. This navigation system accurately re-established mechanical axis without increasing surgical time.

Effect of various hallux valgus reconstruction on sesamoid location: a radiographic study.

BACKGROUND: The correction of sesamoid subluxation is an important component of hallux valgus reconstruction with some surgeons feeling that the sesamoids can be pulled back under the first metatarsal head when imbricating the medial capsule during surgery. The purpose of this study was to radiographically assess the effect of an osteotomy on sesamoid location relative to the second metatarsal. METHODS: This is a retrospective radiographic study review of 165 patients with hallux valgus treated with reconstructive osteotomies. Patients were included if they underwent a scarf or basilar osteotomy for hallux valgus but were excluded if they had inflammatory arthropathy or lesser metatarsal osteotomy. A modified McBride soft tissue procedure was performed in conjunction with the basilar and scarf osteotomies. Each patient's preoperative and postoperative radiographs were evaluated for hallux valgus angle, intermetatarsal 1-2 angle, tibial sesamoid classification, and lateral sesamoid location relative to the second metatarsal. RESULT: The greatest correction of both hallux valgus and intermetatrsal 1-2 angle was achieved in basilar osteotomies (20.6 degrees and 9.7 degrees, respectively), then scarf osteotomies (14.4 degrees and 8.7 degrees, respectively). Basilar and scarf osteotomies both corrected medial sesamoid subluxation relative to the first metatarsal head an average of 2-3 classification stages. All osteotomies had minimal lateral sesamoid location change relative to the second metatarsal. CONCLUSION: The majority of sesamoid correction correlated with the intermetatarsal 1-2 correction. The concept that medial capsular plication pulls the sesamoids beneath the first metatarsal (ie, changes the location of the sesamoids relative to the second metatarsal) was not supported by our results. LEVEL OF EVIDENCE: Level III, retrospective case series.