Case intake and case processing are the most labour-intense aspects of pharmacovigilance and, thus, the most targeted processes for automation and AI solutions. Most of the major players in pharmacovigilance talk about their own solutions or services in pharmacovigilance automation for case processing.

However, we feel like it is still just that: a lot of talk but very little or no real action. That’s why we decided to provide the below summary of tools that are available for everyone and can be used to automate case intake, processing, and reporting.

Innovation driven by costs

The cost of a process or a technology is and should always be an important aspect when discussing pharmacovigilance activities. The safety and risks related to medicine could in theory be predicted, monitored, and reported in infinite detail, if money wasn’t an issue. But money of course is an issue and always will be. Process efficacy is therefore crucial and focus on not wasting time should be much higher than it in fact is.

Automation in pharmacovigilance has been a hot topic for a while now and for a good reason: the price of pharmacovigilance activities has been steadily increasing for the past 10 years and majority of that cost increase is due to the growth of the number of case reports. Also, the nature of case processing work makes you almost immediately want to find an alternative solution for the manual work.

Artificial intelligence, robot process automation and pharmacovigilance software in general have been the buzz words in the pharmacovigilance automation discussion. The key sub-processes to automate is case intake and case processing are receiving, confirming, tracking, reconciling, and forwarding reports of adverse events.

Different companies have approached these two processes in different ways, but the key points are

• automation of data transfer from the source report to a case processing system
• natural language processing (NLP) to analyse the data in the source report
• automated reporting and reconciliation

Intelligent Automation Opportunities in Pharmacovigilance Solutions

As one source to use, TransCelerate Biopharma, a consortium of pharmaceutical companies, has identified the key process steps that could and should be automated and they have published information on “Intelligent Automation Opportunities in Pharmacovigilance Solutions“, including Interactive ICSR and Automation Technologies Tool (IATT).

The tool is split into three main steps:

1. Case Intake
2. Case Processing
3. Case Reporting

TransCelerate has further split these three steps into smaller pieces that can be automated. We look at these steps and the different solutions that can be used to automate a single step or several steps of the process. Also, we will demonstrate how you can skip several of these steps altogether with truly intelligent automation.

Data input unification

In theory, if you can unify the way the case information is received, the rest of the process is relatively easy to automate. Today, the ADRs are traditionally communicated via simple emails (with pre-defined headers in the email body and attachment files), hand filled CIOMS forms which can be printed on one side of a A4 paper, or E2B R2/R3 XML files. In the CIOMS form there are 30+ pre-defined fields that can be filled in. E2B R2 or R3 ICSR is more complicated XML based form but still, the fields are pre-defined.

So, if you can receive the information about the adverse event in a formatted way, it is easy to implement a simple automation that copies the data under the headers and takes systems data (sender and receiver email addresses, system time and date, location, organization etc.) and copies that information into the form fields of a safety database, to the case tracker and to forward the information to the human (pharmacovigilance specialist) who will analyze the data.

The real-world challenge is that usually the AE reports are not received in a pre-defined format, some data can be missing or the CIOMS form is filled in with a hand-writing that is illegible even to the most advanced interpreters of the physician hieroglyphs. The reports are received from doctors and patients during discussions between sales rep and practitioner or medical information specialist and patient asking for help with their medication.

The medical information specialist would probably then transcribe the information into the agreed format but the overqualified sales rep driving around the region visiting clinics probably would send just a quick email from their phone to the PV department. The report could also be received as an email from the patient who is unaware of your format-driven automation, or it can be found in a medical journal or social media site.

So, in practice you can unify how some of the case information is received, but you will never be able to fit all the sources of information into one model. Next, we will look at tools and automation already available for capturing and re-formatting unstructured safety information.

Speech recognition and speech-to-text for case intake

You know how the airlines and mobile phone carriers record the customer service calls “for improving their service”? Same kind of recording could be applied for the medical information phone lines and even the discussions between medical advisors and specialist doctors could be recorded with a phone to start the automation of the process right from the beginning.

If the patient or doctor starts to discuss an adverse event, the company representative who remembers the key points from their annual pharmacovigilance training, would say the magic words “Okay, that sounds like an adverse event, let me stop our discussion for a minute and ask you a few questions.” And then they would proceed asking at least the following questions:

1. “Who is the patient?”
2. “What medical product they were using?”
3. “What was the adverse event they experienced?”

The fourth question, reporter, would already be recorded in the beginning of the discussion.

The speech recognition software (speech-to-text) or the NLP (text-to-data) can be taught to spot these questions and to categorize them as an adverse event report. Once certain bits of the recording have been tagged as an ADR, the speech-to-text is used to transcribe those bits and adverse event report form can be filled and sent automatically to the pharmacovigilance department or the ADR reporting system directly.

This technology is now largely available and there is hardly any reason not to use it. We think companies should adopt it and do so in a smart way. We have noticed some attempts to use speech-to-text technology for case intake but so far it completely misses the point as it’s used for cases already taken in and transcribed from med info or other sources. The whole case intake step can be skipped and traditional “case intake tools” should be made obsolete.

Automatic calling for follow-ups

You can automate your follow-up calls with an automatic call-back service. Once your trained AI or the pharmacovigilance specialist determines which information is missing from the valid ICSR, they mark it as missing in the case processing system and a phone call is placed to the reporter contact details. If phone number is not available, an email is sent to the reporter.

If the reporter answers their phone, the automatic call gives the reporter a chance to either record their answers to the follow-up questions or to connect them to a call with a member of your safety team. The content of the call is recorded, transcribed to text, and analysed for content the same way as described above.

Text-to-data with Natural Language Processing

Once the text is in a text format, it can be analyzed. If the report is already under pre-defined headers thanks to the speech recognition, the accuracy of the NLP is going to be good. If the report is coming from an unstructured email, or a journal article, or a Facebook post or message, the text can be further analyzed by the NLP to recognize keywords and analyze their connections to determine if the text contains safety relevant information.

There are NLP solutions that have been pre-trained to recognize pharmacovigilance related information, or you can choose any solution available and let your pharmacovigilance team train the AI by following their assessments. This isn’t the hard part.

Workflow Automation Software

Connecting different systems to transfer the data automatically reduces the need to download, export, import, and upload data. It also removes the need for one of the biggest pain-points in pharmacovigilance: reconciliation.

There are several different ways to connect different computerized systems (software) you may already be using. Most software solutions have a set of connection tools, API, that can be used to connect it with other computer systems.

There are several different commercial solutions, such as Integromat and Zapier, that offer an easy way to integrate systems and you can automate steps within Outlook or throughout the Office (SharePoint, Excel, Word etc.) suite using Power Automate. Read more about these tools in our blog.

This type of simple automation tools can be used during the Data Acquisition to automatically track incoming data based on rules such as the recipient or sender email address or keywords found in the email subject. You can automate email responses and forwarding to speed up prioritization and triage.

If you are developing software or building a platform to connect your various systems, you might want to look at Google’s Healthcare API. To ensure compliance and compatibility in the healthcare sphere it can be useful to use connection technology specifically built for this and Google is definitely using the right terms (HL7, DICOM, audit logs, HIPAA etc.) to get us interested.

Translation

For translation of safety reports in the local languages, there are several different options. The most well-known is Google Translate which has worse reputation than it deserves. It’s a completely free translation tool that can handle translations between 100+ languages. Google Translate can be easily implemented on different solutions where it automatically recognizes the source language and translates the text to a desired language. Google Translate also works with audio files so it is possible to transcribe the discussions from medical representatives or from medical information calls.

If you still want to avoid using Google Translate, other companies offer services and tools that might fit your accuracy, specialty vocabulary, price, or software requirements better. The translation tool of your choice can be implemented in all the source information fields so that all non-English text from the reporter is translated to English and the original language recognized. This type of quick automated translation can be used for initial triage of the case. If the case seems to contain safety relevant information, it can be assigned to a native speaker or to a professional translator for a better translation and deeper analysis of the case.

Pharmacovigilance platform

A simple way to automate case intake and case processing steps and related tasks is to use a pharmacovigilance platform. With a specialized pharmacovigilance tool, you can store all your data in one place for further process steps: Follow-ups, PSURs, Signal Detection etc. The only such solution on the current market is the Tepsivo Platform (yes, that’s ours).

When you manage your pharmacovigilance processes through Tepsivo Platform, there are immediate process efficiencies. Some of the steps presented in the TransCelerate graphs clearly refer to the current ways of working: sending reconciliation and notification emails, tracking information in Excel, performing duplicate checks, and archiving PDF copies of emails on SharePoint. With Tepsivo Platform, you can leave all these steps in the past and let the automation take care of it.

Start automating today

Someone would ask why we are talking about “can” and “could” when we talk about these solutions and why are these automations not used. The simple answer is that there is no reason why not to use these tools.

All these automation solutions for pharmacovigilance are already available to you straight out of the box. Some of the automation can even be used for free and all the paid software will quickly pay themselves back with saved resources.

If you do not know where or how to start or you are concerned about validation and privacy issues, contact Tepsivo for free consultation for case intake and case processing automation. We will together assess your situation and plan how to proceed choosing the right pharmacovigilance software for you.