Health Monitoring Framework for Weather Radar Based on Long Short-Term Memory Network with a Real Case Study

The health management of weather radar plays a key role in achieving timely and accurate weather forecasting. The current practice mainly exploits a fixed threshold prespecified for some monitoring parameters for fault detection. This causes abundant false alarms due to the evolving working environments, increasing complexity of the modern weather radar, and the ignorance of the dependencies among monitoring parameters. To address the above issues, we propose a deep learning-based health monitoring framework for weather radar. First, we develop a two-stage approach for problem formulation that address issues of fault scarcity and abundant false fault alarms in processing the databases of monitoring data, fault alarm record, and maintenance records. The temporal evolution of weather radar under healthy conditions is represented by a long short-term memory network (LSTM) model. As such, any anomaly can be identified according to the deviation between the LSTM-based prediction and the actual measurement. Then, construct a health indicator based on the portion of the occurrence of deviation beyond a user-specified threshold within a time window. The proposed framework is demonstrated by a real case study for the Chinese S-band weather radar (CINRAD-SA). The results validate the effectiveness of the proposed framework in providing early fault warnings.

Prospective Monitoring of Newly Marketed Drugs in Frail Older Adults Using Real-World Databases

In recent years several new drugs have been approved for treatment of heart failure and type 2 diabetes. Despite their life-prolonging benefits, uptake of new drugs is often slow among older patients with frailty due to under-representation of frail older adults in pivotal clinical trials and concerns for adverse events. To optimize pharmacotherapy, timely evaluation of the drug benefits and risks is urgently needed. We propose a novel drug monitoring framework that prospectively evaluates the effectiveness and safety of newly marketed drugs for frail and non-frail patients in real-world databases. This framework utilizes a validated claims-based frailty index (CFI) (range: 0-1; frail if ≥0.20) to find early signals for effectiveness and safety of new drugs by updating the analysis at regular intervals as new data become available. In this symposium, we present early results of this prospective monitoring framework for 2 new drug classes using Medicare claims data from the approval date until the end of 2017: 1) angiotensin receptor-neprilysin inhibitor (ARNI) (approved in July 2015) for heart failure with reduced ejection fraction (HFrEF) and 2) sodium-glucose cotransporter-2 inhibitors (SGLT2i) (approved in March 2013) for type 2 diabetes. We first show the uptake of ARNI and SGLT2i over time among the eligible Medicare beneficiaries by clinical characteristics, including frailty. Subsequently we present the results of sequential cohort analysis for the effectiveness and safety results of ARNI and SGLT2i. After these presentations, the panel will discuss the strengths, limitations, and challenges of implementing our monitoring framework in real-world databases.

OP223 A Semi-Automated Process To Monitor The Clinical Development And Regulatory Approval Pathway Of Innovative Medicines

IntroductionEarly identification of innovative medicines is crucial for timely health technology assessment (HTA) and efficient patient access. The National Institute for Health Research Innovation Observatory (NIHRIO) identifies, monitors and notifies key HTA stakeholders in England of ‘technologies’ (innovative medicines) within three to five years of regulatory approval. Increasing numbers of innovative medicines and significant uncertainties in clinical and regulatory pathways are major challenges in the monitoring and notification process. An active monitoring framework using pre-defined predictive criteria has previously been developed. This framework provides a standardized and consistent process, but is highly resource-intensive, requiring manual review of individual records.MethodsUsing the previous active monitoring framework, a scoring matrix was calculated and used to prioritize individual technologies using available data in the NIHRIO database: estimated regulatory timelines, regulatory awards/designations, innovative medicine type (for example gene therapies) and clinical trial phase, completion dates and results. A threshold for automatic and manual reviewing of technologies was developed and tested by NIHRIO analysts.ResultsThe scoring system identified approximately ninety percent of technologies meeting the threshold for semi-automated reviewing. The review period for these technologies are set automatically according to predefined criteria depending on data availability. The review periods are updated automatically until the record reaches the threshold that triggers manual reviewing. The remaining ten percent had estimated regulatory timelines necessitating the need for manual reviewing and early engagement with companies to verify regulatory timelines and/or notify HTA stakeholders.ConclusionsPreliminary analysis indicates that each technology is routinely and automatically updated. The semi-automatic updating represents a significant improvement in the efficiency of the monitoring of the large volume of technologies on the NIHRIO database. Ongoing work is being undertaken to further refine, pilot and test the system.This project is funded by the NIHR [(HSRIC-2016-10009)/Innovation Observatory]. The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care.