At a glance
A biotechnology company wanted to better leverage the wealth of data they had gathered to find new insights to aid in research and development, marketing, and sales performance. The company engaged Credera to design and architect a custom solution that leveraged a pluggable artificial intelligence (AI) natural language processing framework to extract these insights, and a custom web application to make them actionable for the business.
Credera implemented a minimum viable product (MVP) AWS-powered analytics platform that leveraged OpenSearch, Lambda, and Sagemaker to handle user requests and generate analytics visualisations in real time. The solution enabled business users to leverage natural language to query complex data sets and seamlessly generate digestible information.
Getting intuitive insights from a complex set of data sources.
The biotechnology company already had a wealth of data regarding patient experiences, healthcare provider feedback, and clinical trial details, but struggled to leverage this data in a meaningful way to aid in business decisions. Their users consistently reported that existing unstructured data solutions required too much reading to make sense of the data, leading them to skip analysing it at all. The organisation realised they needed to develop a system for using this data to generate important indicators that could inform the business about next best actions.
To address user challenges, Credera designed an AI-powered search experience that would accelerate analyst work by allowing users to make searches in natural language and receive key information and analysis related to their searches in an intuitive and navigable format.
Going from idea to insight.
This design phase involved gathering business requirements, sourcing feedback from users leveraging existing systems, prioritising features, and designing the architecture. Front-end user experiences were created in Figma to demonstrate to key stakeholders what could be expected. The solution was split into data science, back-end development, front-end development, and change management workstreams to allow for steady parallel progress.
By analysing feedback from the future userbase, the Credera team prioritised and streamlined development of the analytics platform to target key high value functionalities. Given the unique domain and use cases of the client, special considerations were made for architecture, user experience, front-end design, and offline data operations to adhere to strict medical data governance.
Once an approach was finalised, Credera began a development phase to build out features. The team used an iterative approach to surface design and functionality decisions to key stakeholders. Credera leveraged AWS to host a serverless framework that allowed for a robust search and analytics platform that could scale up or down given demand. The client’s existing natural language processing (NLP) functionality was extended, operationalised, and integrated within the platform to provide real-time analytics on valuable metrics. The resulting solution was a robust and extensible platform with minimal overhead cost.
Realising the value of existing data.
By enabling non-technical business analysts to quickly query complex data sources with natural language, Credera was able to super charge the analyst experience. Analysts can quickly see NLP statistics and trend data over search results. Furthermore, the iterative nature of the search experience allowed analysts to test and validate hypotheses by refining searches with new terms and filter configurations to target useful segments of ingested data.
Additionally, this platform can serve as the foundation of an extensible analytics platform. Almost all aspects of the data pipelines, back-end processing, and front-end rendering were designed to be extended with new datasets, NLP functionality, LLM modelling, search dynamics, and visualisations. By adopting an extensible approach, the client can ensure future feature development with minimal cost impact.