Problem

The client faced a problem in packing and shipping of hazardous materials. There are thousands of business rules regarding what can and cannot be shipped in the same container.

Solution

Fresh Gravity served as a strategic advisor to reduce the client’s excessive shipping costs. We used advanced machine learning methods to empirically “reverse engineer” packing rules by analyzing past shipping data. We then designed an advanced box-packing algorithm to minimize costs attributable to shipping. In addition, we created a User Interface and API for easy input of Sales Orders. These tools effectively provided the recommended packing configuration and associated costs.

Impact

This implementation helped the client in getting an optimized and reliable cost of shipping, materials and labor that can be quoted to the customer. Based on the success of the implementation, we enhanced the solution further for automated ingestion of purchase orders from external sources that have arbitrary formats.

Problem

The client required an automated and intelligent way of extracting relevant data from multiple data sources. Since the existing solutions in the market lacked the ability to understand entity context, the client’s Pharmacovigilance teams spent a significant amount of time tracking, aggregating and analyzing the incoming data.

Solution

Fresh Gravity improved upon the traditional methods for Named Entity Recognition. Machine Learning was used to gather context for tasks such as structured extraction to grasp new words found in medical language and codes.

Impact

Fresh Gravity developed deep learning solutions for the client to help derive context from the sequence of the words to correctly distinguish “Cancer” from “Cancer Institute”. This helped in increasing the speed of the case intake process and reduced the manual effort needed to perform data entry tasks.

Problem

The client has spent significant amount of time reviewing its PhV database to extract information and assemble causality analysis to determine if their drugs caused patient injuries. Regulators request the client to generate safety reports when there are spikes or concerns triggered by safety cases being reported.

Solution

Fresh Gravity worked with the pharmacovigilance team to document all the business rules used today to extract information from safety cases. Our team also documented all the business rules and used them as a starting point for extracting text. We used word embeddings to identify and extract entities for unstructured data and documented all the business rules and used those rules as a starting point for extracting text.

Impact

Fresh Gravity successfully extracted 11 different entities from ArisG with a high degree of accuracy (in most cases accuracy of extraction was more than 90%). We also automated the DILI Causality Scoring which enabled the client to save ~ 2 FTEs for a period of 2 months each time a regulator triggered this review.

Problem

The use of inconsistent reference data like country codes, specialty codes, site status, etc. across applications resulted in costly manual remediation for analytics, and a lack of trust in data by business. The client wanted a solution which could process, manage, and analyze data from multiple sources and would renew the business’s trust in data.

Solution

Fresh Gravity implemented a Reltio MDM system mastering HCP/Investigator, HCO/Facilities, and Address. In conjunction with Reltio MDM, the RDM system served as the single point of reference across the organization. Reltio RDM was implemented to master several reference data entities and provided the interface to Data Stewards to define specialized references without depending on IT.

Impact

Role-based RDM access provided a strong foundation for secure centralized governance. The implementation also helped in reducing the cost and time of manual interventions. The business user had access to a fully integrated user-friendly UI for Data Stewardship that ensured reduction in operational IT costs and restored trust in data by the business.

Problem

The client faced problems in processing large manual steps that involved sorting, filtering, and curating information that sometimes led to rejection of data and was time consuming. Along with this, identification of entities from paper, which is entirely a manual task, proved to be a time consuming and costly process.

Solution

Fresh Gravity developed a LLAMA-based machine learning model to process PubMed articles that generated a summary for the PubMed article highlighting the extracted entities. The solution identified the required entities in the paper using intelligent algorithm to capture context and dependencies between entities.

Impact

After the implementation the client was able to achieve time efficiency of 94.4% and ~85% accuracy in entity identification of independent entities. The solution generated the summary for the PubMed article and highlighted the criticality of annotated dataset gaps for dependent entities identification.

Problem

The client needed an automated approach to validate pregnancy cases to improve the accuracy of their reporting to regulatory authorities. They used three different fields to manually denote that a case is pregnancy related.

Solution

Fresh Gravity solved the problem by leveraging two Natural Language Processing (NLP) algorithms: TD-IDF and BERT. Our team applied TF-IDF to known pregnancy cases to develop a list of terms that are frequently used. Bidirectional Encoder Representations from Transformers is a technique for NLP that is pretrained by Google. This cutting-edge algorithm is trained to have a deeper sense of context while improving prior NLP models that read text in a single direction.

Impact

This model accurately identified more than 95% of pregnancy cases in production environment also improved the quality of client’s Pharmacovigilance process. The algorithm enabled the client to automatically detect and ensure all cases containing pregnancy-related terms can be marked appropriately.

Problem

The client’s objective was to predict which patients will adhere to their drug regimens, and what success factors/pitfalls influence success. The client had also tried several predictive modeling techniques in the past, but none were able to answer the key questions.

Solution

Fresh Gravity trained the model on key predictive feature data from patients, prescribers, and payers. This modelling approach was the appropriate method of creating an anomaly detection tool because it recognized and learned from the sequential pattern data of patients going through a workflow.

Impact

The solution identified anomalous patients and patient cohorts that are at risk of lapsing from drug adherence. They provided interpretability to the models that inform the pharma company which aspects of the process are the sources of most problems.

Problem

The client released a new product that could increase their customer base by 20x. Their support and sales personnel are localized to each country that they support. The client did not have a customer portal that allowed the customer to manage their account and create/manage orders for themselves and their dependents. The client currently has a customer app but all customer app accounts are not associated with the correct provisioned accounts causing the customer to have duplicate accounts.

Solution

Fresh Gravity worked with the client to build a new integrated Amazon-like customer portal that would allow the customer to manage their account/orders by reducing the need to contact support or sales. Fresh Gravity aligned and merged customer accounts based on country-specific match/merge rules and implemented a real-time solution to associate accounts using the combination of match/merge rules and matching transactional data.

Impact

The new solution is slated to be released in Dec 2024 and would allow the client to grow its customer base with little to no impact on their sales and support teams. The solution would also align the customer app accounts with the correct provisioned accounts giving the customer a complete 360 account experience for themselves and their dependents.

Problem

The absence of an ETL tool for data ingestion led to the use of makeshift Canonical files manually curated by an operational service provider. The operational service provider created input data at a pre-defined frequency from third-party data sources procured by the client. Changes to the business transformation rules were very time-consuming and the system was prone to manual errors. Moreover, the non-hierarchical data model resulted in integrity constraint violations, such as a single Stock Keeping Unit (SKU) being mapped to multiple brands. There was duplication of master data leading to client’s paying increased license costs as Semarchy xDM licensing is based on count of golden data. Additionally, replicating the reporting structure required by downstream systems in the existing Master Data Management (MDM) system meant that any changes to the reporting structure required modifications to the data model in MDM, causing disruptions to exist data and requiring ongoing cleanup and maintenance.

Solution

Fresh Gravity assessed the existing implementation of client and recommended several key improvements. They proposed developing direct data pipelines between third-party source data using ETL tools such as Informatica to eliminate manual canonical file creation thus saving upto 50% time spent on manual data load. To support downstream, we recommended transforming the necessary data into a separate staging layer. In this setup, master data would feed into the staging layer. Master data fed staging layer and staging layer fed data to dashboards. Additionally, Semarchy Enrichers, web hooks, match merge rules were fine tuned for performance improvements and data model was optimized through use of minimal fuzzy entities and elimination of duplication of master data thus optimizing Semarchy license costs.

Impact

Capital expenditure on Semarchy license costs reduced. Earlier Semarchy license was tier 2 (Golden records >500K), Fresh Gravity helped reduce golden records to less than 500K thus downshifting to a lower tier of licensing costs. The turnaround was achieved through architectural revisions to the data model and the elimination of the downstream reporting data warehouse replica originally created in the master data management system.

We eliminated the downstream impact of inaccurate reporting thus enabling client’s C-suite to have an accurate view of the company product sales in different geographies. At the same time, downstream dashboard format changes did not require dashboard structure replicated in MDM anymore. This solution was made scalable for anticipated growth in future data volumes without the need for further IT investments or additional data stewards.

Problem

The client was looking to create a combination model comprising of multiple systems and business units (vaccine and pharma) to capture and manage Product Master data, provide a platform for implementing Data Governance processes, and create the foundation for a modern cloud-based technology platform following IDMP standards within the enterprise landscape.

Solution

Fresh Gravity implemented a Product Master solution that adheres to IDMP (Identification of Medicinal Products) compliance standards. The technology of choice was cloud-based Reltio MDM system. The implementation has helped the business in streamlining the process of managing product data across the R&D business, namely, Pharmaceutical Product, Medicinal Product, Manufactured Item, Formulation along with Substance and Product Family.

This solution ensured that high-quality data was created, updated, and maintained in the Reltio system. The workflows were implemented to manage complex business requirements that ensured review and approval of data through a series of predetermined steps. The entire implementation followed strict GxP processes and procedures.

Impact

As a result of this implementation, a solid foundation providing a 360-degree view of Product Master Data (from Substance to Manufactured Item) and robust data governance to maintain the integrity of data within client’s system was established. This foundation also allowed the business to create a long-term roadmap to incorporate more elements of the drug development value chain within the MDM system.