IPSIndian Journal of Pharmacology
Home  IPS  Feedback Subscribe Top cited articles Login 
Users Online : 1811 
Small font sizeDefault font sizeIncrease font size
Navigate Here
  Search
 
  
Resource Links
   Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
   Article in PDF (366 KB)
   Citation Manager
   Access Statistics
   Reader Comments
   Email Alert *
   Add to My List *
* Registration required (free)

 
In This Article
   Need of Artifici...
   Artificial Intel...
   Benefits of Arti...
  Conclusion
   References

 Article Access Statistics
    Viewed1364    
    Printed51    
    Emailed0    
    PDF Downloaded283    
    Comments [Add]    

Recommend this journal

 


 
 Table of Contents    
EDITORIAL
Year : 2019  |  Volume : 51  |  Issue : 6  |  Page : 373-376
 

Artificial intelligence in pharmacovigilance: Practical utility


1 ADR Monitoring Center, PvPI, Department of Pharmacology, PGIMER, Chandigarh, India
2 Department of Pharmacology, PGIMER, Chandigarh, India

Date of Submission18-Dec-2019
Date of Decision30-Dec-2019
Date of Acceptance30-Dec-2019
Date of Web Publication16-Jan-2020

Correspondence Address:
Prof. Bikash Medhi
Department of Pharmacology, PGIMER, Chandigarh - 160 012
India
Login to access the Email id

Source of Support: None, Conflict of Interest: None


DOI: 10.4103/ijp.IJP_814_19

Rights and Permissions



How to cite this article:
Murali K, Kaur S, Prakash A, Medhi B. Artificial intelligence in pharmacovigilance: Practical utility. Indian J Pharmacol 2019;51:373-6

How to cite this URL:
Murali K, Kaur S, Prakash A, Medhi B. Artificial intelligence in pharmacovigilance: Practical utility. Indian J Pharmacol [serial online] 2019 [cited 2020 Apr 8];51:373-6. Available from: http://www.ijp-online.com/text.asp?2019/51/6/373/276049




The word “Pharmacovigilance” was derived from the Greek literature “pharmakon” (means drug) and the word “vigilare” (means keep watch) in Latin.[1] In 1961, the World Health Organization (WHO) has established the pharmacovigilance (PV) program in response to the thalidomide disaster, for global drug monitoring. PV is the science and activities relating to the detection, assessment, understanding, and prevention of adverse effects of drugs or any other possible drug-related problems.[2] Globally, PV is collecting a huge amount of data daily and it is a challenging task to process these vast collected data. Recently, the program has broadened its concerns by including herbals, traditional and complementary medicines, blood products, medical devices, herb vigilance, hemovigilance, and materiovigilance.[1]

There are different databases used in the PV, such as VigiFlow, VigiBase, VigiAccess, and VigiLyze. The Uppsala Monitoring Center (UMC) was established in Uppsala, Sweden, in 1978 as the WHO collaborating center for international drug monitoring. It operates VigiFlow, VigiBase, and VigiLyze on behalf of the WHO and VigiAccess is publicly accessible. There are some methods also in PV, i.e., VigiGrade, VigiMatch, and VigiRank for case report analysis.[3]

On the other hand, the effectiveness of artificial intelligence (AI) can enable to reduce case processing costs to improve PV activities. AI is a computer science field where the computer is trained to perform tasks that normally require human intelligence. AI is organized differently; it is inbuilt with problem-solving proficiency. AI is very profitable and suitable in the field of life sciences, PV, and medical information.[4]

AI has specific features, such as machine learning (ML) and natural learning processing (NLP). ML techniques analyze the structured data such as imaging and genetic data. Unstructured and free-text form is detected by NLP which is being able to understand and interpret human language.[5]

AI is also being used in medical diagnosis, clinical situation to treat, detect, and predict the result, search engines, etc. It is also used in health-care sector to prevent human health problem with proper monitoring.

AI can use a sophisticated algorithm to extract useful information from the huge amount of healthcare data to improve the data accuracy due to its learning and self-correcting ability.[6]


  Need of Artificial Intelligence in Pharmacovigilance Top


PV was designed mainly for patient safety who are limited exposed to treatment drugs during clinical trials and research. This makes it possible to observe drug profiles for a prolonged period and use. It also includes certain groups such as geriatric population, racial groups, pediatric population and pregnant women, and also the incomplete data from long-term drug exposure makes it mandatory to perform PV studies.[7] The approval of life-saving drugs such as anticancer drugs, antitubercular, and antiretroviral drugs is based on fast track system so these drugs could be easily available for the patients and PV performs the assessment, communication of the risk, and effectiveness of these medications.

In developing countries, PV is still a new concept with low preference. Worldwide the countries are raising the issues in concern for the need of systems to monitor the safety of drug postmarketing.[8] Reporting of adverse drug reaction (ADR) is mainly done through spontaneous reporting or by pharmacoepidemiological methods that use systematic collection and analysis of adverse events (AE) associated with the use of drugs. It is also done by Adverse Drug Reaction Monitoring Centres and marketing authorization holder (MAH) industries to solve emerging problems, record signals, and communicate to minimize or prevent harm.

Nowadays, drug safety is a major threat after launching the new drug to the market. During the clinical trial or after marketing, the major source of erosion is unpredictable toxicities that cause morbidity and mortality from normal dose of the drug. As per the record from 2008 to 2017, the Food and Drug Administration (FDA) approved 321 novel drugs. At the same time FDA, AE Reporting System has reported a total of >10 million AE reports in which 5.8 million were recorded as serious, and 1.1 million were death reports.[9]

ADR data need to be collected by license holder that should be from pharmaceutical company and submitted to the local drug regulatory authority. The most important operation in PV industry is detection and reporting of ADRs, coding of AE in technical terms, preparing safety individual reports, assessment of seriousness, and relationship with suspected drug. All of these depend on the human interference, which is time-consuming; and hence, the detection of ADRs requires a new technology. A multinational pharmaceutical company in collaboration with a professional services company has developed an AI and ML system to facilitate the processing and maintenance of quality and compliance standards. The globally available data are so vast that it cannot be analyzed manually where AI becomes useful to track them. AI techniques play a significant role in the area of drug design and identification of AEs of pharmaceutical products.[9]


  Artificial Intelligence in Pharmacovigilance Top


Individual Case Safety Report (ICSR) is a document in a specific format for recording information with the use of FDA regulations to provide the information on AEs, product-related issues, and consumer reports.[10]

The Central Drugs Standards Control Organization (CDSCO) (Pharmacovigilance Gsr 287 € dated 8-03-2016, REGD.D.L.-33004/99) has made it mandatory for the MAHs to report ICSR of the marketed drug in India to National Coordination Center for Pharmacovigilance Programme of India (NCC-PvPI) as well as to them. Currently, 64,441 ICSR has been collected and submitted to NCC-PvPI, Ghaziabad. Finally, these reports will be sent to WHO-UMC, Sweden, through VigiFlow software.[11] Because of increasing number of ICSR reporting, the handling and the processing of these reports is becoming tedious and time-consuming job including the high cost. Therefore, PvPI may also adopting the Machine Intelligence techniques such as AI for reducing workforce, cost, and time for ICSR case processing.[10]

There are two categories in AI in ICSR processing:

  1. Insertion of structured and unstructured content: insert information through XML, DOCX, Image, and PDF for reading the case. NLP and ML are used to extract ICSR information in a regulatory compliant manner
  2. AI for decision-making: the quality of ICSR is usually poor. Therefore, AI may play an important role in making the unlisted or individual random AEs, drug classifiers, correlation, etc.


AI technologies in PV are very helpful in the extraction of accurate information. AI tools can automate or facilitate almost every aspect of PV in case processing, risk tracking, which reduces the total processing time.[12] These are some tools which are useful in PV activities:

VigiBase

A PV database that records the information in a structured and ordered form to allow easy analysis of recorded data. This system is related to medical and drug classifications, including the WHO-Adverse Reaction Terminology/Medical Dictionary for Regulatory Activities (MedDRA), WHO International Classification of Disease, and WHODrug. Over 20 million reports of adverse drug effect were recorded by VigiBase (as of May 2019).[13]

VigiAccess

It is a publicly accessible web application to browse and access the data of adverse drug effects easily through VigiBase.[13]

VigiLyze

It is an online resource that provides a clear and quick review of VigiBase which can be explored online for further analysis.[14]

VigiFlow

It is a web-based ICSR management system for international drug monitoring by collection, processing, and sharing of data to facilitate effective data analysis, which is supported by WHODrug and MedDRA.[15]

VigiGrade

To measure the completeness score of clinically relevant information in ordered format on the individual case report. This is mainly used as a part of communication with countries on data quality.[16]

VigiMatch

It is an algorithm to detect the similar individual case report by the use of probabilistic pattern matching.[17]

VigiRank

It is a novel method to detect the statistical signals which is not just for disproportionate reporting patterns but also for the completeness, recency, and geographic spread of individual case reporting.[18]

The patient data stored electronically is easy to diagnose AE, with additionally provided information as symptoms of disease, phase, and severity of the disease and contributing factors. These data act as electronic health record and it represents the stored and collected additional information that generates the usefulness of AI.

An important domain of PV is signal detection and its clinical assessment. The WHO-UMC uses the Bayesian Confidence Propagation Neural Network to detect signal in ICSR database. More than a single report is required to detect the signal; it depends on the quality of the information and severity of the events. Data mining software (Empirica Signal, PV-analyzer) is also detect the drug safety issue. In India, CDSCO along with NCC-PvPI regulated this signal generation using the database such as VigiBase. The signals are used by PvPI to review, identify, decide, and conclude the collected data from various national databases.[19]


  Benefits of Artificial Intelligence in Pharmacovigilance Top


  1. The most important benefits of AI are reduced cycle times. Due to this method, the processing is spontaneous
  2. Improve the quality and accuracy of the information
  3. AI can handle or manage diverse types of incoming data formats
  4. It can be used for the identification of ADRs
  5. AI is useful to reduce the burden and time of case processing
  6. AI tools extract the information from the adverse drug event form and evaluate the case validity without the workforce.


There are many applications of AI in PV, and it is bound to have an economic impact on the PV field.


  Conclusion Top


Henceforth, AI techniques will be useful in identifying and initiate a hidden relationship for accurate ICSR processing in PV. Nowadays, awareness about AI in PV is infancy. This awareness can be influenced by the collaboration of IT firms and pharmaceutical companies, through which latter and medical device companies could improve regulatory compliance, achieve cost reduction, etc. Nowadays, there are IT systems that automate the case processing and ADR reporting, but still, all process requires manual effort such as case intake and data entry. The overall process from case receipt to reporting can be automated with the help of AI process. These processes will not only reduce the cost, but also it will improve the quality and accuracy. Until 2017, most of the individuals were not aware about the AI therefore, it is necessary to increase the awareness of AI in PV. Few of web applications of AI are accessible to public like “VigiAccess” for data of ADRs. For PvPI, the burden of the overall process from case receipt to reporting can be reduced by automated input with the help of AI techniques. These processes will not only reduce the cost, also it will improve the quality and accuracy. The automated data are harmonized world widely and UMC Sweden is capable to monitor the collected data by different PV centers.

The future strategies of drug safety could become more advanced, driven by AI techniques. More researches are needed in the field of AI with respect to PV. AI, databases, and tools are in primary stage of development, and it could poof its advancement in future in the field of PV.



 
  References Top

1.
Deye N, Vincent F, Michel P, Ehrmann S, da Silva D, Piagnerelli M, et al. Changes in cardiac arrest patients' temperature management after the 2013 TTM trial: Results from an international survey. Ann Intensive Care 2016;6:4.  Back to cited text no. 1
    
2.
Meyboom RH, Egberts AC, Gribnau FW, Hekster YA. Pharmacovigilance in perspective. Drug Saf 1999;21:429-47.  Back to cited text no. 2
    
3.
UMC WHO [Internet]. Available from: https://www.who-umc.org/about-us/who-we-are/. [Last modified on 2018 Oct 11].  Back to cited text no. 3
    
4.
Barr A, Feigenbaum E, Roads C. The Handbook of Artificial Intelligence, Volume 1. Comput Music J 1982;1:3-14.  Back to cited text no. 4
    
5.
Mockute R, Desai S, Perera S, Assuncao B, Danysz K, Tetarenko N, et al. Artificial Intelligence Within Pharmacovigilance: A Means to Identify Cognitive Services and the Framework for Their Validation. Pharmaceut Med 2019;33:109-20.  Back to cited text no. 5
    
6.
Yu KH, Beam AL, Kohane IS. Artificial intelligence in healthcare. Nat Biomed Eng 2018;2:719-31.  Back to cited text no. 6
    
7.
Council for International Organisations of Medical Sciences. Benefit-risk balance for Marketed drugs: evaluating safety signals. Report of CIOMS Working Group IV. Geneva, CIOMS, 1998. No. 762.  Back to cited text no. 7
    
8.
Venulet J, Helling-Borda M. WHO's international drug monitoring – The formative years, 1968-1975: Preparatory, pilot and early operational phases. Drug Saf 2010;33:e1-23.  Back to cited text no. 8
    
9.
Basile AO, Yahi A, Tatonetti NP. Artificial intelligence for drug toxicity and safety. Trends Pharmacol Sci 2019;40:624-35.  Back to cited text no. 9
    
10.
Available from: https://www.fda.gov/industry/fda-resources-data-standards/individual-case-safety-reports. [Last accessed on 2018 Mar 27].  Back to cited text no. 10
    
11.
12.
White RW, Wang S, Pant A, Harpaz R, Shukla P, Sun W, et al. Early identification of adverse drug reactions from search log data. J Biomed Inform 2016;59:42-8.  Back to cited text no. 12
    
13.
Shankar PR. VigiAccess: Promoting public access to VigiBase. Indian J Pharmacol 2016;48:606-7.  Back to cited text no. 13
[PUBMED]  [Full text]  
14.
Bhushan S, Ray RS, Prakash J, Singh GN. Global versus Indian Perspective of Pioglitazone-induced Adverse Drug Reactions Including Bladder Cancer: A Comparative Retrospective Pharmacovigilance Analysis. Clin Ther 2019;41:2252-62.  Back to cited text no. 14
    
15.
Plöen M, Wallberg M, Olsson S. User-driven development of a web-based tool for patient reporting of drug-related harm. Drug Saf 2015;38:359-64.  Back to cited text no. 15
    
16.
Bergvall T, Norén GN, Lindquist M. vigiGrade: A tool to identify well-documented individual case reports and highlight systematic data quality issues. Drug Saf 2014;37:65-77.  Back to cited text no. 16
    
17.
Tregunno PM, Fink DB, Fernandez-Fernandez C, Lázaro-Bengoa E, Norén GN. Performance of probabilistic method to detect duplicate individual case safety reports. Drug Saf 2014;37:249-58.  Back to cited text no. 17
    
18.
Caster O, Sandberg L, Bergvall T, Watson S, Norén GN. vigiRank for statistical signal detection in pharmacovigilance:First results from prospective real-world use. Pharmacoepidemiol Drug Saf 2017;26:1006-10.  Back to cited text no. 18
    
19.
Ventola CL. Big Data and Pharmacovigilance: Data Mining for Adverse Drug Events and Interactions.P T 2018;43:340-51.  Back to cited text no. 19
    




 

Top
Print this article  Email this article
 

    

Site Map | Home | Contact Us | Feedback | Copyright and Disclaimer
Online since 20th July '04
Published by Wolters Kluwer - Medknow