i2e Consulting - An ML-Driven Process for Drug Order and Route Selection to Optimize Stock, Cost, and Resources


Sri Kanagala, with:

  • Reshma Katkar
  • Nitesh Singh
  • Sri Harsha Akavarapu
  • Azeem Shaikh

Country: United States

Organization: i2e Consulting

Formed in 2008, i2e Consulting is a pioneering software services company specializing in delivering tailored technology solutions and consultancy services to organizations within the pharmaceutical industry. Recognized as one of the fastest-growing private companies in the US in the 2018 Inc. 5000, i2e Consulting has become synonymous with innovation and the digitization of the healthcare sector.

With a focus on Data Analytics, Project Portfolio Management, Automation, and Cloud services, i2e Consulting offers a range of customizable solutions that drive portfolio decisions down to operational execution. The company's commitment to continuous improvement of process and data security is evidenced by its ISO9001 and ISM27001 certifications.

Central to i2e Consulting's innovative approach is its strategic partnership with Dataiku, a collaboration that leverages the platform's AI and Machine Learning capabilities. This partnership enables i2e Consulting to resolve complex problems and provide optimized solutions, positioning them at the forefront of technological advancement.

Further strengthening its position as a thought leader, i2e Consulting has developed a robust partnership ecosystem with leading technology partners, including Planisware, a provider of project portfolio management (PPM) solutions. This collaboration with Planisware, along with partnerships with Microsoft, Snowflake, and AWS, enhances i2e Consulting's ability to integrate effective solutions tailored to unique requirements.

Named as a leader in the field, i2e Consulting's innovative approach, strategic alliances, and commitment to excellence have enabled them to provide value to organizations of all sizes, principally within the pharmaceutical sector. Their dedication to client success and technological innovation continues to set them apart as a sophisticated and forward-thinking company in the industry.

Awards Categories:

  • Best Partner Acceleration Use Case
  • Best Moonshot Use Case
  • Best MLOps Use Case


Business Challenge:

A global pharma giant was struggling with manual processes to predict the drug dosage inventory needed for clinical trials and other drug studies at various sites. Each site had a team of supply chain leads (SCLs) who were responsible for the drug inventory management. The team used to manually determine and order the next dosage of drugs which then needed to be shipped from the nearest depot. 

The SCLs were spending considerable time navigating through multiple databases, and Spotfire dashboards to collect the dosage inventory, number of ongoing subjects for the study, etc., and then manually calculating the inventory of dosages. 

Key Challenges :

  1. The manual process was tedious, time consuming and error prone, hence it resulted in:
  2. Ordering excess stock which was expiring,
  3. Cost escalations,  
  4. Drug sitting idle in one clinical site,
  5. Mismatches between drug expiration and transit timelines.




Business Solution:

Dataiku provides built-in connectors which facilitated collection of data from multiple sources and load them in a centralized warehouse for further processing.

Exploratory data analysis was performed on the collated data which comprise of Inventory information, Active subjects at each depot and site and historical transfer routes information, etc. by leveraging the various available visual recipes such as Prepare, Join, Group, Filter, etc.

Next, we created and trained ML models using Dataiku platform. Dataiku supports good MLOps practices so we can seamlessly deploy, monitor, and manage machine learning projects.

Data science experts at i2e also trained the model to recommend:

  • The best available routes,
  • The time taken for the drug to get shipped to the site,
  • Demand forecasting for upcoming requirements,
  • Screening success rate for subjects based on historical clinical trial conversion data for any clinical study.

While predicting, the system was made agile enough to take into consideration complex scenarios which could affect the dosage, for example, change in the number of subjects, visits required per subject, expiry dates of the drugs etc. We trained around 15+ machine learning algorithms which included Logistic regression, Random Forest, Decision tree, etc. We captured the metric values for all these machine learning models, and Random Forest model gave best results on the available data.

The above steps are implemented by creating Dataiku Flow and Dataiku Scenario with time-based triggers to automate the entire process. On successful run of scenario, the latest recommendation is shared with end users via email notification. In addition, the Dataiku API endpoint takes input from end users via external systems like Spotfire, etc. And triggers the Dataiku Scenario on demand to facilitate a near real-time recommendation.

The provided solution helped the supply chain leads to make an informed decision as to when to place the order and which route would be most viable considering the drugs expiration date and transfer metrics information.




Day-to-day Change:

Earlier Supply chain leads (SCL) were collecting inventory data from multiple tools, downloading those reports and manually collating them to performs calculations in an Excel sheet for drug requirements of the next 6/12 months based on the clinical study plan and then manually place the order for multiple depot and sites.

The proposed solution eliminated the risk of human errors in calculations and provided the below mentioned benefits:

  • Accurate predictions eliminated the risk of ordering excess drug doses. 
  • The SCLs have one central system to get a bird’s eye view through a Spotfire dashboard. 
  • Manual work of collating the information and calculating the drug dosage is eliminated. 
  • The best available transportation routes helped the SCLs to take an informed decision. 
  • Predictions are now based on the transit time vs the expiry dates of drugs. 

Business Area Enhanced: Supply Chain/Supplier Management/Service Delivery

Use Case Stage: In Production


Value Generated:

The results obtained from implementation of above solution provided a seamless interaction between various data sources and generated dependable insights for the SCL’s:

  • 95% increase in accuracy of drug dosage recommendations.
  • 10 hours per week saved by the SCLs which were otherwise spent on manual work.
  • 30% of drug costs saved due to accurate drug predictions
  • Significant reduction in mismatches between drug expiration and transit timelines 
  • Developed an intelligent order system which recommends quantity to be ordered from the best possible routes with ability to expedite the order based on urgent requirements in the depot and sites


Value Brought by Dataiku:

Dataiku provides support for a wide range of connectors which facilitates collection of data from multiple sources.

Exploratory analysis can be performed easily on the collated data with the help of visual recipes and plugins. The preview feature to perform transformation on the sample data and observe the output enables quick data preparation and feature extraction.

Both coders and non-coders can make use of these visual recipes and perform exploratory data analysis and can also train Machine learning models supported by Dataiku.

Dataiku also has a feature to automate the created pipeline and notify stakeholders about its success or failure.

Improvements in the current process using Dataiku:

  • Dataiku provides features like in-build connectors for connecting to different databases/sources, deployment capabilities, supports to explainable AI, including interactive reports for feature importance, partial dependence plots, etc. This features not only saves time of developers for solution implementation but also helped SCL’s and business for fast data driven decision making and helped them to place the orders in a short span of time
  • With a wide range of machine learning and analytic tasks of Dataiku, such as prediction, clustering, time series, image classification and much more. We used this feature in the use case and through which SCL’s were able to predict study screening conversion accurately and provided recommendations for drug requirements of upcoming months with good accuracy which helped business to avoid excess ordering of drugs and expiration of drugs at depot/site - which in turns saved huge amount of cost involved in drug manufacture, transfer and shipments
  • For security concerns, SSO authentication and role-based access control are set up in Dataiku to make sure only the legitimate users with required permissions should be able to access the data. We were able to leverage this functionality in the use case to maintain role based access for developers and business.

Value Type:

  • Reduce cost
  • Save time

Value Range: Hundreds of thousands of $


A great use case with significant potential: I like how easily data was consolidated to train and test various ML models within a single workflow, allowing for the selection of the best-performing model. The ability to do all of this within a single platform, without the need to manage multiple tools, along with their associated costs and different standards, is heavily underrated.

Version history
Publication date:
09-08-2023 04:10 PM
Version history
Last update:
‎08-10-2023 05:38 PM
Updated by: