Pharmacovigilance: AI to fortify Post-Market Drug Quality in Pharma

Original article was published by Anu Ganesan on Artificial Intelligence on Medium


Pharmacovigilance: AI to fortify Post-Market Drug Quality in Pharma

Pharmacovigilance, also referred to as PV or PhV refers to the drug safety process that involves collecting, analyzing, and communicating the adverse effect of drugs. Pharmacological science also involves medication errors such as overdosing or abusing drugs following worldwide laws and regulations related to drugs.

Adverse Event Reporting

The players involved in reporting adverse events of any drugs are the drug safety department in the pharmaceutical companies and the government-based drug regulatory authorities. The adverse event report of any drugs can come from different sources like healthcare professionals, patients, patient support group, clinical reports or post-market studies, drug regulatory authorities, literature sources, the report made in social media, news publication, etc.,

The various phases in adverse event reporting include receipt, triage, data entry, assessment, report. Adverse Event Reporting is regulatory requirements and companies or hospitals failing to do so will be fined accordingly.

A valid Individual Case Safety Report (ICSR) is identified by the below characteristics:

– Identifiable Patient

– Identifiable Reporter

– Suspected Drug

– Adverse Event

Sometimes the ICSR reportings are not straight-forward and vary based on countries, healthcare professionals, and hospitals. Validating ICSR is a mundane process that also involves judgment calls.

Coding of Adverse Event

The Adverse Event Reports about the drugs are not easy to interpret as reporting varies based on the individuals reporting and composing the reports. Such reports are coded using standardized terminology from a medical code dictionary like MedDRA. For instance, headache can be coded as PT Headache where PT stands for Preferred Term. Once the adverse effect of drugs is coded and standardized, it becomes easier to analyze and report the findings in common language.

An adverse event is termed serious if it involves death, in-patient hospital visits, life-threatening medical conditions, and birth defects.

Adverse Event Risk Management

Risk Management involves signal detection based on causality assessment on aggregated reports and confirming or refuting adverse events.

Signal Detection

Once the adverse event is reported and standardized, it undergoes a review phase where the signals are validated. As of 2019, there are more than 120 countries registering their ICSR reports in Vigibase which is a global ICSR database managed by WHO. All these ICSR reports come from pharmaceutical companies, drug regulatory authorities and large healthcare providers.

Aggregate Reporting

Aggregate reporting refers to the periodic reporting of all the adverse events for a given drug. Any drugs in the pre-market are reported in NDA Annual Reports, Clinical Study Reports (CSR), and IND Annual Reports. Any drugs in the post-market phase are reported in Periodic Safety Update Report (PSUR), Summary Bridging Report (SBR), Development Safety Update Report (DSUR), Annual Safety Report (ASR), Periodic Adverse Drug Experience Report (PADER)

Causality Assessment

Determining the cause for the adverse event is one of the challenging problems in Pharmacovigilance. It needs judgment calls and gathering the necessary information to relate the drug to the adverse event.

The Common Terminology Criteria for Adverse Event (CTCAE) has many grade levels ranging from grade 1 to grade 5 based on the seriousness of the report. The causality can be re-challenged in which case the entire workflow for pharmacovigilance should be revisited. There are measures to validate the causality of adverse events by making sure the report was made within 15 days of witnessing possible drug side-effects.

Confirm / Refute Adverse Event

The global database like Vigibase storing all the drug effects along with the availability of a wide range of data sources for pharma-related information along with technological advancement in computing space has made it possible for mining pharma data on large scale.

The detected signals from adverse event reporting are assessed and validated by statistically measuring the threshold and monitoring for adverse events violating the threshold. Once the adverse events are confirmed, it is mandatory to report to all the drug agency in different countries like the FDA, Food and Drug Administration

Conclusion

AI in Pharmacovigilance

The reporting of any drugs from either patients, healthcare professionals, or other related entities can be submitted via email, phone, or fax. All these adverse event reports can also come from EHR (Electronic Health Record) logged by physicians, image processing records, and medical publications. Due to the adverse nature of data sources across different countries, there is a threat to the quality of pharmacovigilance data.

Post-market monitoring of drugs provides more information than included in the clinical trials like high-risk groups, long term effects, food and drug interaction, increased severity of known and unknown side-effects over time. Employing AI in post-market monitoring of drugs improves compliance and reduces cost of processing each and every ICSRs. Any non-serious ICSR reporting should be processed via a touch-less case processing AI system where there is no manual intervention needed. AI simplifies the signal detection phase by applying narrative analysis to extract case and aggregate similar case reports for a given drug.

The mundane nature of filing ICSR reports leads to human mistakes in capturing all the necessary information when reports come in as drug/dosage inquiry, safety precaution, or as an adverse effect of drug usage. AI-adopted pharmacovigilance assists not only in assessing the quality of the reported events but also classify the adverse nature of such events. AI along with Augmented Intelligence and a wide range of medical data from different parts of the world is paving way for unified global healthcare.

Follow Us for more information on Machine Learning Platforms and Solutions

Predera LinkedIn

Predera