A pharmaceutical company faces fundamental inefficiencies in determining the production costs of new or modified medicines. Estimates are carried out manually through the finance department, take months, and hinder decision-making regarding development, pricing, and market launch. Additionally, the process is prone to inaccuracies because it cannot flexibly reflect changes in composition, production processes, or input prices. The result is delayed innovation, higher planning costs, and the risk of poor investment decisions.
COGS (Cost of Goods)
Client NameZentiva
Client CountryCZ
- Client typeEnterprise
- IndustryHealthcare, Pharma & Biotech
- Application areasFinance & Accounting, Product Development, Innovation, R&D, Strategy, Planning & Decision-Making
- AI technologiesAdvanced Analytics / Data Science, Explainable & Trustworthy AI, Machine Learning, Simulation / Digital Twins
- Business impactsData-Driven Decisions & Planning, Employee Enablement & Productivity
- Data typesStructured Tabular Data
- Delivery modelsConsulting, Custom Development, Product / Licensed Software
- DeploymentsCloud
- Key capabilitiesPlanning, Scheduling & Optimization, Predictive Analytics & Forecasting
- Project stagesInitial Production Deployment
- Solution formsOther
Solution Description
Problem description
Solution
The solution utilizes a data model trained on historical drug manufacturing costs, enabling fast and accurate price predictions for new or modified products. The model accounts for changes in composition, manufacturing processes, materials, capacities, and logistics. Users enter product parameters and immediately receive a detailed cost estimate, including breakdowns. The system significantly reduces calculation time from months to minutes, increases accuracy, and enables rapid scenario analysis for strategic decision-making.
Main Users of the Solution
Technologists, Production, Portfolio managers
Project timeframe (months)
2 years
Technologies used
Frontend Application: ReactJS Backend Application: Python + GraphQL API Machine Learning: Sklearn, Pytorch, LightGBM Database: PostgreSQL + DuckDB (analytical database)
Additional services
- AI strategy and roadmap
- Audit / feasibility study
- Identification and prioritization of suitable use cases
- Data collection and preprocessing
- Annotation / Synthetic data / Dataset expansion
- Data governance and data quality
- AI model selection and customization
- Change management support and user training
- Systematic AI education programs
- Ensuring MLOps infrastructure
- Ongoing maintenance and model retraining
Implementation
Project Owner on the Client's Side
IT / Data / Technology Lead
Participation on the Client's Side
- Business / Product Owner
- Domain / Process Experts
- Project and Change Management
- Quality, Security, Compliance
Form of Supplier Involvement
Complete implementation
Operation and Maintenance
Operational Model
Internal team + DevOps together
Needed Competencies on the Client's Side
People responsible for the drug pricing process, technologists, production. DevOps.
Other Resources or Infrastructure
Azure
Impact and Results
Qualitative Benefits
Estimation of the total price of any drug modification, from months to an instant result. The original long duration was caused by the fact that several departments participated in the estimation of the total price of the drug and it was done manually. In the new state, the algorithm contains the logic for determining the price for all departments.
Quantitative Results
Ability to instantly obtain an accurate estimate of production costs. Ability to estimate the business case for the production cost of new pharmaceuticals. Possibly adjusting existing ones to reduce production costs.
Client Feedback
Positive, the project is constantly developing and being integrated so that more departments have access.
Lessons Learned and Recommendations
Key Success Factors
Deep understanding of the pharmaceutical manufacturing process and its costs. Not naively applying AI and ML methods as a black box, but applying methods at the lowest level. Continuous evaluation of results with the client. Excellent guidance from the client, providing information on pricing methods from key personnel for individual steps of the production process.
Biggest Challenges
Studying the entire production cost calculation process in detail. Data acquisition and integration into the AI algorithm.
Recommendation for Others
All pricing logic in digital and verifiable form. Estimates available immediately. It must be carried out with a team of people who understand the production cost process in depth.
Promotion
Demo / Public Resources
- Demo přístupu na nagenerovaných datech k dispozici v případě zájmu.
- CompanyDNAI
- ContactJakub Szasz
- Emailjakub.szasz@dnai.ai
- Websitehttps://www.dnai.ai
- AddressU Nikolajky 3, 150 00 Praha