A Comparative Effectiveness Study on Opioid Use Disorder Prediction Using Artificial Intelligence and Existing Risk Models.

Opioid use disorder (OUD) is a leading cause of death in the United States placing a tremendous burden on patients, their families, and health care systems. Artificial intelligence (AI) can be harnessed with available healthcare data to produce automated OUD prediction tools. In this retrospective study, we developed AI based models for OUD prediction and showed that AI can predict OUD more effectively than existing clinical tools including the unweighted opioid risk tool (ORT). Data include 474,208 patients' data over 10 years; 269,748 were females with an average age of 56.78 years. Cases are prescription opioid users with at least one diagnosis of OUD or at least one prescription for buprenorphine or methadone. Controls are prescription opioid users with no OUD diagnoses or buprenorphine or methadone prescriptions. On 100 randomly selected test sets including 47,396 patients, our proposed transformer-based AI model can predict OUD more efficiently (AUC = 0.742 ± 0.021) compared to logistic regression (AUC = 0.651 ± 0.025), random forest (AUC = 0.679 ± 0.026), xgboost (AUC = 0.690 ± 0.027), long short-term memory model (AUC = 0.706 ± 0.026), transformer (AUC = 0.725 ± 0.024), and unweighted ORT model (AUC = 0.559 ± 0.025). Our results show that embedding AI algorithms into clinical care may assist clinicians in risk stratification and management of patients receiving opioid therapy.

A human-derived neurovascular unit in vitro model to study the effects of cellular cross-talk and soluble factors on barrier integrity.

The blood-brain barrier (BBB) restricts paracellular and transcellular diffusion of compounds and is part of a dynamic multicellular structure known as the "neurovascular unit" (NVU), which strictly regulates the brain homeostasis and microenvironment. Several neuropathological conditions (e.g., Parkinson's disease and Alzheimer's disease), are associated with BBB impairment yet the exact underlying pathophysiological mechanisms remain unclear. In total, 90% of drugs that pass animal testing fail human clinical trials, in part due to inter-species discrepancies. Thus, in vitro human-based models of the NVU are essential to better understand BBB mechanisms; connecting its dysfunction to neuropathological conditions for more effective and improved therapeutic treatments. Herein, we developed a biomimetic tri-culture NVU in vitro model consisting of 3 human-derived cell lines: human cerebral micro-vascular endothelial cells (hCMEC/D3), human 1321N1 (astrocyte) cells, and human SH-SY5Y neuroblastoma cells. The cells were grown in Transwell hanging inserts in a variety of configurations and the optimal setup was found to be the comprehensive tri-culture model, where endothelial cells express typical markers of the BBB and contribute to enhancing neural cell viability and neurite outgrowth. The tri-culture configuration was found to exhibit the highest transendothelial electrical resistance (TEER), suggesting that the cross-talk between astrocytes and neurons provides an important contribution to barrier integrity. Lastly, the model was validated upon exposure to several soluble factors [e.g., Lipopolysaccharides (LPS), sodium butyrate (NaB), and retinoic acid (RA)] known to affect BBB permeability and integrity. This in vitro biological model can be considered as a highly biomimetic recapitulation of the human NVU aiming to unravel brain pathophysiology mechanisms as well as improve testing and delivery of therapeutics.

Mapping the mAb Aggregation Propensity Using Self-Interaction Chromatography as a Screening Tool.

The osmotic second virial coefficient ( B2), which describes protein-protein molecular interactions in solution, was determined using self-interaction chromatography (SIC) for an IgG1-type mAb across a wide range of solution conditions. These data were compared with its time dependent aggregation behavior, as determined using size-exclusion chromatography (SEC), and its temperature dependent aggregation behavior using dynamic light scattering (DLS) over a four-week period (SEC) or overnight (DLS). DLS and SEC gave consistent data on aggregation behavior, which correlated well with experimental B2 trends across the wide pH (4-9) and NaCl concentration (0-1.0 M) ranges studied. The IgG aggregated at pH 4 for 0.5-1.0 M NaCl concentrations and for 0 M NaCl concentrations at pH 8. Best stability against aggregation was exhibited for the pH range from 5 to 8 at 0.8-1.0 M NaCl. SIC data were able to be classified within the one-day solution conditions for aggregation, which were not identified for 2-3 weeks in the accelerated SEC stability study. The ability of SIC to provide such data rapidly reflects the fundamentally thermodynamic nature of this parameter and of the aggregation process itself. Proteins with attractive protein-protein interactions and negative B2 coefficients in the range -3 to -6 clearly exhibit aggregation behavior, while B2 values in the range 0 to 2 showed good stability toward aggregation. SIC allows the rapid screening of solution conditions for which mAbs will exhibit stability to aggregation while requiring 90% less time and material compared with that required for a conventional SEC aggregation study.