Model-Agnostic Meta-Learning-based Region-Adaptive Parameter Adjustment Scheme for Influenza Forecasting

Deep learning models perform well when there is enough data available for training, but otherwise the performance deteriorates rapidly owing to the so-called data shortage problem. Recently, model-agnostic meta-learning (MAML) was proposed to alleviate this problem by embedding common prior knowledge from different tasks into the initial parameters of the target model. Data shortages are very common in regional influenza predictions, and MAML also often struggles with regional influenza forecasting, especially when region-specific knowledge, such as peak timing or intensity, varies. In this paper, we propose a novel MAML-based parameter adjustment scheme for influenza forecasting, called MARAPAS. The fundamental idea of our scheme is to adjust the initial parameters obtained from common knowledge to a target region by using adjustment variables. We experimentally show that MARAPAS outperforms other MAML-based methods, in terms of root mean square error and Pearson correlation coefficient. Particularly, this scheme improves the forecasting performance by up to 34 % compared with that of the state-of-the-art schemes. We also show the robust forecasting accuracy of our scheme and demonstrate its applicability by performing zero-shot COVID-19 forecasting.

A randomized, double-blind, Phase 1 study of IN-006, an inhaled antibody treatment for COVID-19 (preprint)

Rationale: Although COVID-19 is predominantly a respiratory tract infection, current antibody treatments are administered by systemic dosing. We hypothesize that inhaled delivery of a muco-trapping monoclonal antibody would provide a more effective and convenient treatment for COVID-19. Objective: We investigated the safety, tolerability, and pharmacokinetics of IN-006, a reformulation of regdanvimab, an approved intravenous treatment for COVID-19, for nebulized delivery by a handheld nebulizer. Methods: A Phase 1 study was conducted in healthy volunteers. Study staff and participants were blinded to treatment assignment, except for pharmacy staff preparing the study drug. The primary outcomes were safety and tolerability. Exploratory outcomes were pharmacokinetic measurements of IN-006 in nasal fluid and serum. Results: Twenty-three participants were enrolled and randomized across two single dose and one multiple dose cohorts. There were no serious adverse events (SAEs). All enrolled participants completed the study without treatment interruption or discontinuation. All treatment-emergent adverse events were transient, non-dose dependent, and were graded mild to moderate in severity. Nebulization was well tolerated and completed in a mean of 6 minutes in the high dose group. Mean nasal fluid concentrations of IN-006 in the multiple dose cohort were 921 microgram/gram of nasal fluid at 30 minutes after dosing and 5.4 microgram/gram at 22 hours. Mean serum levels in the multiple dose cohort peaked at 0.55 microgram/mL at 3 days after the final dose. Conclusions IN-006 was well-tolerated and achieved concentrations in the respiratory tract orders of magnitude above its inhibitory concentration. These data support further clinical development of IN-006.

Stable nebulization and muco-trapping properties of Regdanvimab/IN-006 support its development as a potent, dose-saving inhaled therapy for COVID-19 (preprint)

The respiratory tract represents the key target for antiviral delivery in early interventions to prevent severe COVID-19. While neutralizing monoclonal antibodies (mAb) possess considerable efficacy, their current reliance on parenteral dosing necessitates very large doses and places a substantial burden on the healthcare system. In contrast, direct inhaled delivery of mAb therapeutics offers the convenience of self-dosing at home, as well as much more efficient mAb delivery to the respiratory tract. Here, building on our previous discovery of Fc-mucin interactions crosslinking viruses to mucins, we showed that regdanvimab, a potent neutralizing mAb already approved for COVID-19 in several countries around the world, can effectively trap SARS-CoV-2 virus-like-particles in fresh human airway mucus. IN-006, a reformulation of Regdanvimab, was stably nebulized across a wide range of concentrations, with no loss of activity and no formation of aggregates. Finally, nebulized delivery of IN-006 resulted in 100-fold greater mAb levels in the lungs of rats compared to serum, in marked contrast to intravenously dosed mAbs. These results not only support our current efforts to evaluate the safety and efficacy of IN-006 in clinical trials, but more broadly substantiate nebulized delivery of human antiviral mAbs as a new paradigm in treating SARS-CoV-2 and other respiratory pathologies.

Transformer-Based Models for Question Answering on COVID19 (preprint)

In response to the Kaggle's COVID-19 Open Research Dataset (CORD-19) challenge, we have proposed three transformer-based question-answering systems using BERT, ALBERT, and T5 models. Since the CORD-19 dataset is unlabeled, we have evaluated the question-answering models' performance on two labeled questions answers datasets \textemdash CovidQA and CovidGQA. The BERT-based QA system achieved the highest F1 score (26.32), while the ALBERT-based QA system achieved the highest Exact Match (13.04). However, numerous challenges are associated with developing high-performance question-answering systems for the ongoing COVID-19 pandemic and future pandemics. At the end of this paper, we discuss these challenges and suggest potential solutions to address them.