Transforming ICSR Reporting: Reducing Case Processing Time by 40% and Boosting Productivity with Gen AI Automation
Project Overview
Our client, a leading pharmaceutical company, faced challenges with their manual Individual Case Safety Report (ICSR) tool. The existing process was prone to errors, time-consuming, and impacted regulatory compliance. To address these issues, we collaborated with the client to automate the reporting process using Generative AI. Our goal was to streamline workflows, reduce human involvement, and enhance user experience without disrupting established processes.
Impact
40%
Improved Accuracy
1.5x
Enhanced Productivity
80%
Reduced Manual Interaction
100%
On-Time Report
17%
Reduced Operational Costs
Dashboard of their cases categorized by priorities
Translation page where users can see the cases being translated from Original to English
Translation page where users can see the cases being translated from Original to English
Introduction:
Why are we doing this?
Our client, a leading pharmaceutical company, faced significant challenges with their manual Individual Case Safety Reporting (ICSR) process. The existing system struggled with low accuracy, inefficient workflows, and high operational costs. These issues led to frequent errors, delays in reporting, and compliance risks. To address these pain points, the client sought a solution that could streamline their workflow, enhance productivity, and ensure accuracy through the integration of Generative AI (Gen AI).
Key Challenges
Manual Process Time
Each case took around 45 minutes to process.
Accuracy Rate
Only 85%, leading to frequent errors.
Productivity
Limited to 10 reports per user per day.
Discovery Phase:
Understanding User and Business Needs
We started by analyzing the existing ICSR process to find key challenges. We discovered issues like complex manual data entry, repetitive tasks causing user frustration, and errors leading to compliance risks. This understanding helped us design a solution that would simplify workflows and improve efficiency.
Our Key Findings:
Error Rate
35% of cases required rework due to manual errors, increasing operational costs.
User Frustration
The repetitive nature of data entry led to burnout and mistakes, affecting employee retention.
Inefficient Workflow
Manual processes slowed down case handling, reducing overall productivity and impacting service delivery timelines.
Business Impact
Inefficiencies in the system led to higher costs, reduced scalability, and limited the ability to meet growing market demands.
Conducting workshop with users to understand needs and build user stories
Images showing user stories captured from workshops with users and business people
Defining the Solution:
Building a Gen AI Workflow
We mapped out a new automated workflow that integrated Gen AI capabilities. The goal was to retain familiar elements from the existing system to ensure a smooth transition for users.
Major Goals:
Automated Data Extraction
35% of cases required rework due to manual errors, increasing operational costs.
Reduced Human Interaction
Simplified processes with automation, minimizing the need for manual involvement and reducing user frustration.
Increased Efficiency
Automated workflows improved case handling speed, enhancing overall productivity.
Enhanced Scalability
Optimized processes support higher case volumes, improving business efficiency and reducing operational costs.
Conducting workshop with users to refine the solution we built
Aligning user stories
Ideation and Design
As the sole designer, I created wireframes and high-fidelity prototypes, ensuring they aligned with user needs and technical feasibility.
Key Design Decisions:
User-Friendly Interface
We simplified the interface design to make it intuitive and easy to navigate, ensuring a smoother user experience.
Transparent Processes
We provided clear insights into automated decision-making, building user trust and confidence in the system.
Stakeholder Collaboration
We actively incorporated feedback from stakeholders throughout the design process, ensuring the solution aligned with both user needs and business goals.
Some Rough sketches and wireframes on identifying teh layout and designs
Dashboard
The new dashboard has significantly improved user efficiency and accuracy. By quickly highlighting high-priority cases, task completion time has reduced by 30%, and missed deadlines have decreased by 25%. The streamlined interface makes information easily accessible, boosting productivity by 40%. Additionally, real-time updates provide immediate insights, enhancing case-handling accuracy by 20%.
Key Outcomes:
Task Categorization
Easy Identification
User Efficiency
Easily Monitor Case Workload
Main Dashboard Screen
Cases filtered in Dashboard
Case Translation
The case translation page allows users to view the original document alongside its English translation, with key entities highlighted for easy reference. Users can edit the translation directly, ensuring better accuracy. This setup has reduced translation errors by 30% and increased user satisfaction by 20%, improving clarity and comprehension.
Key Impact:
Side-by-Side View
Entity Highlighting
Editable Translations
Enhanced Clarity
Case Translated screen with Source Language and Translated Language
Acknowledgement Screen for Case Translation
Entity Extraction
The entity extraction feature automatically identifies and highlights key entities like names, dates, and medical terms within the document. Users can easily verify and edit the extracted entities to ensure accuracy, leading to a 30% reduction in manual corrections. This streamlined workflow reduces manual effort by 40%, speeds up data processing, and boosts overall productivity by 25%.
Key Impact:
Easily Identifying Narratives
Faster Decision Making
Boosting Productivity
Business Efficiency
Entity coding showcasing extracted entities from med dictionary
Entity extracted from Source Document
Case Validation
Automated validity checks streamline the case validation process, reducing manual effort and enabling users to validate cases with minimal input. AI-driven duplicate and seriousness checks enhance workflow efficiency by minimizing human interaction, while reducing validation time by 30% for faster case processing. Additionally, automated validation improves accuracy by 20%, reducing errors and ensuring reliable case handling.
Key Impact:
Easy Validation
Less Human Interaction
Faster Turnaround
Increased Accuracy
Validating case through given criteria
Once the case is validated
Narrative Generation
The narrative generation automatically compiles extracted entities and key information into a summarized report, ready for submission. This process reduces manual effort by 40%, cuts errors by 30%, and speeds up report creation by 50%, ensuring faster, more accurate submissions with minimal user input.
Key Impact:
Faster Report Creation
Quickly Generating Conclusions
Easily Editable
Increased Efficiency
Final Generated Narrative
Editing the Generated Narrative
Outcomes
Task completion time reduced by 30%, enabling faster prioritization of high-priority cases.
Error reduction by 25%, minimizing missed deadlines and improving case handling accuracy.
User productivity boosted by 40% through a streamlined interface, reducing time spent searching for information.
The project was a huge success, leading to the client's satisfaction and the acquisition of additional projects.
Learnings
Owning the entire design process as the sole designer was challenging, requiring adaptability and focus.
Collaborating closely with developers and data scientists helped deepen my understanding of LLMs and their real-world applications.
The cross-functional collaboration enhanced my problem-solving skills and fostered a collaborative mindset for future projects.
Gained valuable insights into how design can effectively complement complex systems like AI and data science.
Transforming ICSR Reporting: Reducing Case Processing Time by 40% and Boosting Productivity with Gen AI Automation
Project Overview
Our client, a leading pharmaceutical company, faced challenges with their manual Individual Case Safety Report (ICSR) tool. The existing process was prone to errors, time-consuming, and impacted regulatory compliance. To address these issues, we collaborated with the client to automate the reporting process using Generative AI. Our goal was to streamline workflows, reduce human involvement, and enhance user experience without disrupting established processes.
Client Impact
40%
Improved Accuracy
1.5x
Enhanced Productivity
80%
Reduced Manual Interaction
100%
On-Time Report
17%
Reduced Operational Costs
Impact
Improved Accuracy
Automated data extraction reduced errors by 40%, significantly decreasing the need for rework.
Strengthened Compliance
Built-in compliance checks ensured 100% on-time reporting, minimizing regulatory risks.
Increased User Satisfaction
User-reported frustration decreased by 50% due to a simplified interface and reduced manual interaction.
Business Efficiency
The optimized system increased case processing capacity by 60%, supporting higher volumes and reducing operational costs.
Enhanced Productivity
Streamlined workflows reduced case handling time by 30%, allowing users to manage cases more efficiently.
Dashboard of their cases categorized by priorities
Translation page where users can see the cases being translated from Original to English
Translation page where users can see the cases being translated from Original to English
Introduction:
Why are we doing this?
Our client, a leading pharmaceutical company, faced significant challenges with their manual Individual Case Safety Reporting (ICSR) process. The existing system struggled with low accuracy, inefficient workflows, and high operational costs. These issues led to frequent errors, delays in reporting, and compliance risks. To address these pain points, the client sought a solution that could streamline their workflow, enhance productivity, and ensure accuracy through the integration of Generative AI (Gen AI).
Key Challenges
Manual Process Time
Each case took around 45 minutes to process.
Accuracy Rate
Only 85%, leading to frequent errors.
Productivity
Limited to 10 reports per user per day.
Discovery Phase:
Understanding User and Business Needs
We started by analyzing the existing ICSR process to find key challenges. We discovered issues like complex manual data entry, repetitive tasks causing user frustration, and errors leading to compliance risks. This understanding helped us design a solution that would simplify workflows and improve efficiency.
Our Key Findings:
Error Rate
35% of cases required rework due to manual errors, increasing operational costs.
User Frustration
The repetitive nature of data entry led to burnout and mistakes, affecting employee retention.
Inefficient Workflow
Manual processes slowed down case handling, reducing overall productivity and impacting service delivery timelines.
Business Impact
Inefficiencies in the system led to higher costs, reduced scalability, and limited the ability to meet growing market demands.
Conducting workshop with users to understand needs and build user stories
Images showing user stories captured from workshops with users and business people
Defining the Solution:
Building a Gen AI Workflow
We mapped out a new automated workflow that integrated Gen AI capabilities. The goal was to retain familiar elements from the existing system to ensure a smooth transition for users.
Major Goals:
Automated Data Extraction
35% of cases required rework due to manual errors, increasing operational costs.
Reduced Human Interaction
Simplified processes with automation, minimizing the need for manual involvement and reducing user frustration.
Increased Efficiency
Automated workflows improved case handling speed, enhancing overall productivity.
Enhanced Scalability
Optimized processes support higher case volumes, improving business efficiency and reducing operational costs.
Conducting workshop with users to refine the solution we built
Aligning user stories
Ideation and Design
As the sole designer, I created wireframes and high-fidelity prototypes, ensuring they aligned with user needs and technical feasibility.
Key Design Decisions:
User-Friendly Interface
We simplified the interface design to make it intuitive and easy to navigate, ensuring a smoother user experience.
Transparent Processes
We provided clear insights into automated decision-making, building user trust and confidence in the system.
Stakeholder Collaboration
We actively incorporated feedback from stakeholders throughout the design process, ensuring the solution aligned with both user needs and business goals.
Some Rough sketches and wireframes on identifying teh layout and designs
Dashboard
The new dashboard has significantly improved user efficiency and accuracy. By quickly highlighting high-priority cases, task completion time has reduced by 30%, and missed deadlines have decreased by 25%. The streamlined interface makes information easily accessible, boosting productivity by 40%. Additionally, real-time updates provide immediate insights, enhancing case-handling accuracy by 20%.
Key Outcomes:
Task Categorization
Easy Identification
User Efficiency
Easily Monitor Case Workload
Main Dashboard Screen
Cases filtered in Dashboard
Case Translation
The case translation page allows users to view the original document alongside its English translation, with key entities highlighted for easy reference. Users can edit the translation directly, ensuring better accuracy. This setup has reduced translation errors by 30% and increased user satisfaction by 20%, improving clarity and comprehension.
Key Impact:
Side-by-Side View
Entity Highlighting
Editable Translations
Enhanced Clarity
Case Translated screen with Source Language and Translated Language
Acknowledgement Screen for Case Translation
Entity Extraction
The entity extraction feature automatically identifies and highlights key entities like names, dates, and medical terms within the document. Users can easily verify and edit the extracted entities to ensure accuracy, leading to a 30% reduction in manual corrections. This streamlined workflow reduces manual effort by 40%, speeds up data processing, and boosts overall productivity by 25%.
Key Impact:
Easily Identifying Narratives
Faster Decision Making
Boosting Productivity
Business Efficiency
Entity coding showcasing extracted entities from med dictionary
Entity extracted from Source Document
Case Validation
Automated validity checks streamline the case validation process, reducing manual effort and enabling users to validate cases with minimal input. AI-driven duplicate and seriousness checks enhance workflow efficiency by minimizing human interaction, while reducing validation time by 30% for faster case processing. Additionally, automated validation improves accuracy by 20%, reducing errors and ensuring reliable case handling.
Key Impact:
Easy Validation
Less Human Interaction
Faster Turnaround
Increased Accuracy
Validating case through given criteria
Once the case is validated
Narrative Generation
The narrative generation automatically compiles extracted entities and key information into a summarized report, ready for submission. This process reduces manual effort by 40%, cuts errors by 30%, and speeds up report creation by 50%, ensuring faster, more accurate submissions with minimal user input.
Key Impact:
Faster Report Creation
Quickly Generating Conclusions
Easily Editable
Increased Efficiency
Final Generated Narrative
Editing the Generated Narrative
Outcomes
Task completion time reduced by 30%, enabling faster prioritization of high-priority cases.
Error reduction by 25%, minimizing missed deadlines and improving case handling accuracy.
User productivity boosted by 40% through a streamlined interface, reducing time spent searching for information.
The project was a huge success, leading to the client's satisfaction and the acquisition of additional projects.
Learnings
Owning the entire design process as the sole designer was challenging, requiring adaptability and focus.
Collaborating closely with developers and data scientists helped deepen my understanding of LLMs and their real-world applications.
The cross-functional collaboration enhanced my problem-solving skills and fostered a collaborative mindset for future projects.
Gained valuable insights into how design can effectively complement complex systems like AI and data science.
Transforming ICSR Reporting: Reducing Case Processing Time by 40% and Boosting Productivity with Gen AI Automation
Project Overview
Our client, a leading pharmaceutical company, faced challenges with their manual Individual Case Safety Report (ICSR) tool. The existing process was prone to errors, time-consuming, and impacted regulatory compliance. To address these issues, we collaborated with the client to automate the reporting process using Generative AI. Our goal was to streamline workflows, reduce human involvement, and enhance user experience without disrupting established processes.
Impact
40%
Improved Accuracy
1.5x
Enhanced Productivity
80%
Reduced Manual Interaction
100%
On-Time Report
17%
Reduced Operational Costs
Dashboard of their cases categorized by priorities
Translation page where users can see the cases being translated from Original to English
Translation page where users can see the cases being translated from Original to English
Introduction:
Why are we doing this?
Our client, a leading pharmaceutical company, faced significant challenges with their manual Individual Case Safety Reporting (ICSR) process. The existing system struggled with low accuracy, inefficient workflows, and high operational costs. These issues led to frequent errors, delays in reporting, and compliance risks. To address these pain points, the client sought a solution that could streamline their workflow, enhance productivity, and ensure accuracy through the integration of Generative AI (Gen AI).
Key Challenges
Manual Process Time
Each case took around 45 minutes to process.
Accuracy Rate
Only 85%, leading to frequent errors.
Productivity
Limited to 10 reports per user per day.
Discovery Phase:
Understanding User and Business Needs
We started by analyzing the existing ICSR process to find key challenges. We discovered issues like complex manual data entry, repetitive tasks causing user frustration, and errors leading to compliance risks. This understanding helped us design a solution that would simplify workflows and improve efficiency.
Our Key Findings:
Error Rate
35% of cases required rework due to manual errors, increasing operational costs.
User Frustration
The repetitive nature of data entry led to burnout and mistakes, affecting employee retention.
Inefficient Workflow
Manual processes slowed down case handling, reducing overall productivity and impacting service delivery timelines.
Business Impact
Inefficiencies in the system led to higher costs, reduced scalability, and limited the ability to meet growing market demands.
Conducting workshop with users to understand needs and build user stories
Images showing user stories captured from workshops with users and business people
Defining the Solution:
Building a Gen AI Workflow
We mapped out a new automated workflow that integrated Gen AI capabilities. The goal was to retain familiar elements from the existing system to ensure a smooth transition for users.
Major Goals:
Automated Data Extraction
35% of cases required rework due to manual errors, increasing operational costs.
Reduced Human Interaction
Simplified processes with automation, minimizing the need for manual involvement and reducing user frustration.
Increased Efficiency
Automated workflows improved case handling speed, enhancing overall productivity.
Enhanced Scalability
Optimized processes support higher case volumes, improving business efficiency and reducing operational costs.
Conducting workshop with users to refine the solution we built
Aligning user stories
Ideation and Design
As the sole designer, I created wireframes and high-fidelity prototypes, ensuring they aligned with user needs and technical feasibility.
Key Design Decisions:
User-Friendly Interface
We simplified the interface design to make it intuitive and easy to navigate, ensuring a smoother user experience.
Transparent Processes
We provided clear insights into automated decision-making, building user trust and confidence in the system.
Stakeholder Collaboration
We actively incorporated feedback from stakeholders throughout the design process, ensuring the solution aligned with both user needs and business goals.
Some Rough sketches and wireframes on identifying teh layout and designs
Dashboard
The new dashboard has significantly improved user efficiency and accuracy. By quickly highlighting high-priority cases, task completion time has reduced by 30%, and missed deadlines have decreased by 25%. The streamlined interface makes information easily accessible, boosting productivity by 40%. Additionally, real-time updates provide immediate insights, enhancing case-handling accuracy by 20%.
Key Outcomes:
Task Categorization
Easy Identification
User Efficiency
Easily Monitor Case Workload
Main Dashboard Screen
Cases filtered in Dashboard
Case Translation
The case translation page allows users to view the original document alongside its English translation, with key entities highlighted for easy reference. Users can edit the translation directly, ensuring better accuracy. This setup has reduced translation errors by 30% and increased user satisfaction by 20%, improving clarity and comprehension.
Key Impact:
Side-by-Side View
Entity Highlighting
Editable Translations
Enhanced Clarity
Case Translated screen with Source Language and Translated Language
Acknowledgement Screen for Case Translation
Entity Extraction
The entity extraction feature automatically identifies and highlights key entities like names, dates, and medical terms within the document. Users can easily verify and edit the extracted entities to ensure accuracy, leading to a 30% reduction in manual corrections. This streamlined workflow reduces manual effort by 40%, speeds up data processing, and boosts overall productivity by 25%.
Key Impact:
Easily Identifying Narratives
Faster Decision Making
Boosting Productivity
Business Efficiency
Entity coding showcasing extracted entities from med dictionary
Entity extracted from Source Document
Case Validation
Automated validity checks streamline the case validation process, reducing manual effort and enabling users to validate cases with minimal input. AI-driven duplicate and seriousness checks enhance workflow efficiency by minimizing human interaction, while reducing validation time by 30% for faster case processing. Additionally, automated validation improves accuracy by 20%, reducing errors and ensuring reliable case handling.
Key Impact:
Easy Validation
Less Human Interaction
Faster Turnaround
Increased Accuracy
Validating case through given criteria
Once the case is validated
Narrative Generation
The narrative generation automatically compiles extracted entities and key information into a summarized report, ready for submission. This process reduces manual effort by 40%, cuts errors by 30%, and speeds up report creation by 50%, ensuring faster, more accurate submissions with minimal user input.
Key Impact:
Faster Report Creation
Quickly Generating Conclusions
Easily Editable
Increased Efficiency
Final Generated Narrative
Editing the Generated Narrative
Outcomes
Task completion time reduced by 30%, enabling faster prioritization of high-priority cases.
Error reduction by 25%, minimizing missed deadlines and improving case handling accuracy.
User productivity boosted by 40% through a streamlined interface, reducing time spent searching for information.
The project was a huge success, leading to the client's satisfaction and the acquisition of additional projects.
Learnings
Owning the entire design process as the sole designer was challenging, requiring adaptability and focus.
Collaborating closely with developers and data scientists helped deepen my understanding of LLMs and their real-world applications.
The cross-functional collaboration enhanced my problem-solving skills and fostered a collaborative mindset for future projects.
Gained valuable insights into how design can effectively complement complex systems like AI and data science.