Leveraging Generative AI in PBI for Pharma: Case Studies

It is important to note that currently the pharmaceutical industry is in the process of getting digitalized than before following the demands towards increased fast drug development, patients’ care, and operation optimization. At the same time, here one can discuss the following perspective in this regard – the integration of Generative AI with Power BI. Deep learning AI, one of the more modern divisions of artificial intelligence, which aims at attaining the existing data patterns and utilizing them to produce new data, can be extremely valuable in anticipating the outcomes, simulating biological processes, and even identifying new drugs. With the integration of such applications, APQC’s study reveals that this technology has the potential for giving the pharma firms an advantage in data leveraging, if it is combined with Power BI’s robust data visualization and data analysis tools.

It can help Pharma companies to perform a better forecast for outcomes and improve the drug discovery and development process, adopt Power BI, an easy platform for managing supply chains along with patient-centric treatments, integrate proper pharmacovigilance for safer medicines. Not only does the big data enhances complex strategies, but also it also improves the procedural decision making on research, policies, and healthcare delivery as the researchers, policy makers, and healthcare professionals are provided with more meaningful data to provide better health solutions.

We have taken few Case Study to look into the aspect of empowering Pharma with Generative AI-Driven Insights through Power BI :-

Optimizing Drug Supply Chain Management

Objective: Large pharma company needed to optimize its supply chain to ensure timely delivery of vaccines and medications, reduce costs, and minimize waste. The complexity of global logistics, coupled with fluctuating demand and strict regulatory requirements, made this a formidable challenge.

Solution: It integrated Generative AI with Power BI to enhance supply chain management. Generative AI algorithms analyzed historical data, current supply chain status, and market conditions to predict future demand and supply needs. Power BI dashboards visualized these insights, offering real-time, actionable information.

Identifying the Objectives:

Pfizer first identified critical pain points within their supply chain:

• Inaccurate demand forecasting
• High inventory costs
• Delays in delivery
• Inefficiencies in responding to market fluctuations

The primary objectives were to improve forecast accuracy, reduce costs, and enhance on-time delivery rates.

Data Collection and Integration:

Pfizer collected extensive data from various sources:

• Historical sales data
• Current inventory levels
• Supplier information
• Market trends and demand forecasts
• Production schedules
• Regulatory requirements

This data was integrated into a centralized data warehouse to ensure consistency and accessibility.

Deploying Generative AI Models:

Generative AI models were developed and deployed to analyze the integrated data. These models used machine learning algorithms to:

• Predict future demand for different products.
• Optimize inventory levels by determining the best stock levels to maintain.
• Identify potential supply chain disruptions and suggest mitigation strategies.

?Integration with Power BI:

The insights generated by the AI models were fed into Power BI. This involved:

• Creating data pipelines that regularly updated Power BI with the latest predictions and insights from the AI models
• Developing interactive dashboards and reports to visualize key metrics, such as predicted demand, inventory levels, and delivery performance.

?Customizing Dashboards for Stakeholders:

Different dashboards were tailored for various stakeholders:

• Supply Chain Managers: Focused on operational metrics, including inventory levels, order fulfillment rates, and supplier performance.
• Executives: Highlighted strategic insights, such as cost savings, overall supply chain efficiency, and high-level demand forecasts.
• Logistics Teams: Provided real-time tracking of shipments and predictive analytics for potential delays or issues.

Training and Change Management:

To ensure effective use of the new system, Pfizer implemented a comprehensive training program:

• Workshops and Tutorials
• User Guides and Support

?Impact and Results:

• Improved Forecast Accuracy: The AI models significantly enhanced the accuracy of demand forecasts, reducing overstock and stockouts. Improve forecast accuracy by 20%
• Cost Reduction: Optimized inventory levels led to a 15% reduction in holding costs.
• Enhanced Delivery Rates: Improved logistics planning resulted in a 25% increase in on-time deliveries.
• Faster Decision-Making: Real-time, data-driven insights enabled quicker and more informed decision-making.

Improving Patient Outcomes with Personalized Medicine

Objective: Pharma company sought to improve patient outcomes by personalizing treatment plans based on individual patient data, including genetic information, lifestyle factors, and treatment history.

Solution: Novartis leveraged Generative AI to analyze vast amounts of patient data and generate personalized treatment recommendations. These recommendations were visualized in Power BI dashboards, making it easier for healthcare providers to understand and utilize the information.

Data Collection and Integration - Consolidating Patient Data

• Details: Novartis began by collecting vast amounts of patient data from various sources, including electronic health records (EHRs), genetic information, treatment history, lifestyle factors, and real-time health monitoring devices.
• Tools: EHR systems, genomic sequencing platforms, wearable health tech, and patient surveys.

?Data Preprocessing - Cleaning and Structuring Data

• Details: The collected data was often unstructured and heterogeneous. Data scientists at Novartis employed data preprocessing techniques to clean, normalize, and structure the data for analysis.
• Tools: ETL (Extract, Transform, Load) processes, data cleaning tools, and data warehouses like SQL databases.

Building the Generative AI Models - Developing and Training AI Models

• Details: Novartis developed Generative AI models capable of analyzing patient data to identify patterns and predict the effectiveness of various treatment options. These models were trained using machine learning algorithms on historical data to learn relationships between patient characteristics and treatment outcomes.
• Tools: Python (with libraries like TensorFlow, PyTorch), Jupyter Notebooks, and cloud-based AI platforms like Azure Machine Learning or AWS SageMaker.

Connecting AI Insights to Power BI

• Details: The insights generated by the AI models were integrated into Power BI. This involved setting up data pipelines to transfer AI outputs into Power BI datasets, where they could be visualized and interacted with.
• Tools: Power BI Desktop, Power BI Service, APIs for data integration, and data connectors.

Designing User-Friendly Interactive Dashboards

• Details: Data analysts at Novartis designed Power BI dashboards to present AI-generated insights in an intuitive and actionable manner. These dashboards included personalized treatment recommendations, visualizations of predicted treatment outcomes, and comparative analyses of different treatment options.
• Tools: Power BI Desktop for dashboard creation, DAX (Data Analysis Expressions) for complex calculations, and Power Query for data transformation.

Outcomes and Benefits

• Improved Patient Outcomes: Personalized treatment plans led to a 20% increase in treatment efficacy.
• Enhanced Patient Satisfaction: Patients received care tailored to their individual needs, improving satisfaction and adherence to treatment plans.
• Resource Optimization: The solution reduced unnecessary treatments and optimized the use of medical resources, lowering overall healthcare costs.

Key Technologies and Tools Used

• Data Collection: EHR systems, genomic sequencing platforms, wearable devices.
• Data Preprocessing: ETL tools, SQL databases.
• AI Model Development: Python, TensorFlow, PyTorch, cloud-based AI platforms.
• Data Integration and Visualization: Power BI Desktop, Power BI Service, APIs.
• User Trining and Deployment: Training programs, Power BI Service.

About the Author:

Anshuman Dubey is a seasoned Senior Business Consultant at Infosys Consulting with more than 16 years of extensive proficiency in data products, Data-Mart, data governance, data modeling, security, data visualization, and data consulting. He boasts a proven track record of executing mission-critical projects across diverse facets of the life sciences and healthcare sectors, spanning commercial and operational domains. Anshuman excels in both agile methodologies and waterfall approaches, demonstrating effective collaboration with cross-functional teams throughout his career.