Savana
Unlocking Real-World Evidence with AI-Driven Approach

Research indicates that approximately 80 percent of pertinent clinical information, including symptoms and disease diagnoses, resides within the free-text, ‘unstructured’ component of electronic health records (EHR). This has led Savana, an international medical company, to develop a scientific methodology that uses artificial intelligence (AI) to unleash the full clinical value embedded within EHRs’ free-text documentation.

The firm leverages computational linguistics to extract, transform and assess data from various electronic medical records (EMR) systems. AI allows for a comprehensive inclusion of all patients’ information pertaining to a specific condition, eliminating selection bias. In addition, Savana relies on researchers’ expertise to create a gold standard, which serves as a benchmark for training its AI models. Known as the world’s largest AI-enabled, multilingual and multi-center research network, the firm produces high-quality deep realworld evidence (RWE) studies that lead to enhanced patient outcomes.

“We have prepared diligently for this moment, and our efforts are geared towards enhancing the use of AI in healthcare by efficiently extracting valuable information from medical records to promote research, patient care and evidence-based decision-making,” says Ignacio H. Medrano MD, MSc, founder and chief medical officer of Savana.

The company distinguishes itself from the prevailing emphasis on AI applications in diagnostic tools, mammograms, CT scans and predictive models by automating database creation using multilingual clinical notes sourced from diverse sites. Savana automates the collection of real-world evidence by leveraging the capabilities of machine learning (ML) and natural language processing (NLP).

The database incorporates information from various medical specialties, such as oncology, internal medicine and the emergency room. This results in the reduction of the comprehensive approach, selection and follow-up biases, resulting in higher-quality and more comprehensive data.

Savana embraces an open-source paradigm to disseminate its methodology in the esteemed Journal of Medical Informatics. The methodology involves researchers annotating data, a process akin to captcha verification on websites. The firm equips these researchers with a sophisticated tool that discerns potential medical conditions from text fragments. The iterative confirmation and refutation of findings by the researchers effectively train the machine, ensuring remarkable precision in their publications and engendering trust from prominent journals and regulatory authorities. It is Savana’s unique focus on a scientific methodology complementing its technology that sets them apart from its competitors.

Savana has successfully collaborated with over 250 sites spanning 12 to 13 countries across Europe and the USA. The company has expanded its reach to five languages and partnered with major pharmaceutical and mid-sized biotech organizations. Savana’s success lies in its ability to provide a compelling narrative to hospitals, healthcare providers and sites, emphasizing the untapped potential of existing information and urging them to consider the benefits of data reuse through appropriate technology.

The firm also prioritizes data privacy and security, proactively de-identifying and anonymizing information. Complying with European regulations, it advocates for Institutional Review Board (IRB) protocols before research data utilization. To overcome this, it has to have face-to-face conversations with healthcare providers’ managers to enable the reuse of hospitals’ information for research through AI.

A case study that further illustrates the application of top AI techniques in analyzing clinical data is when a pharmaceutical company was engaged in research related to rheumatoid arthritis and pulmonary fibrosis, with a focus on understanding drug interactions. They sought to determine the extent to which patients with one condition also experienced the other. This was required in record time and without creating a new registry or observational study. Savana leveraged its advanced epidemiological and descriptive capabilities and technology and rapidly identified and quantified the number of individuals with specific conditions.

Due to its versatility in addressing diverse evidence-generation needs and business requirements, Savana is wellequipped to adapt its approach accordingly.