INTERVIEW: Can AI Cut the Cost of Clinical Trials Without Compromising Data Quality?

South African health technology company BioInformatiCo is using artificial intelligence to automate clinical trial database preparation, after its Leova platform scored high at the 2026 CDISC AI Innovation Challenge. The company says Leova cut database testing and validation from five days of specialist work to one in a Phase 2 HIV vaccine study, as funding pressures threaten medical research programmes. In this interview with Impact Newswire, founder and CEO Shirley Collie discusses how the technology works, its limits, BioInformatiCo’s plans to raise $2.3 million and whether African-developed AI can make clinical research faster, more affordable and more accessible.

Clinical trials can take years and cost millions of dollars to bring a new medicine or vaccine from development to market, making delays and inefficiencies in research infrastructure a significant challenge for drug developers. Before a trial can begin enrolling participants, research teams must translate study protocols into reliable systems for collecting, validating and managing clinical data. Artificial intelligence is increasingly being explored as a way to automate parts of this process, reduce manual workloads and improve consistency, although regulatory compliance, data integrity and human oversight remain essential.

A South African-developed platform is seeking to address this challenge. Johannesburg-based clinical data management company BioInformatiCo’s Leova Synthetic Data Generator placed second in Use Case 1 of the 2026 CDISC AI Innovation Challenge, which focused on AI-enabled synthetic data generation for software testing. The competition attracted 50 submissions from 30 organisations and individual entrants across three use cases, with Leova recognised for its approach to translating study protocols into a standardised model and generating repeatable synthetic-data scenarios to test clinical study builds.

The technology operates in a sector where standardisation and traceability are critical. The Clinical Data Interchange Standards Consortium (CDISC) develops data standards used to structure and exchange clinical research information, while the US Food and Drug Administration requires supported data standards for many electronic submissions involving drug and biological product studies. These frameworks help regulators review clinical evidence consistently, making the quality of the underlying data systems an important part of the research process.

Leova’s development also comes amid financial pressure on parts of South Africa’s medical research ecosystem. In 2025, US funding suspensions and grant terminations disrupted health programmes and threatened clinical research, including HIV and tuberculosis studies. Médecins Sans Frontières and Treatment Action Group identified 39 South African HIV and TB clinical research sites at risk, with at least 27 HIV trials and 20 TB trials potentially affected. The disruptions have highlighted the vulnerability of research programmes that depend on international funding and the need to make available resources go further.

BioInformatiCo says Leova can reduce the time needed to test and validate a clinical trial database. In one study cited by the company, the process fell from five days of specialist work to one. The platform is designed to complement, rather than replace, clinical data management expertise, with human oversight helping maintain the reliability and traceability required in clinical research. Its international recognition raises broader questions about the role African-developed AI can play in improving research efficiency, strengthening local technical capacity and supporting clinical studies that address diseases affecting African populations.

In this interview with Impact Newswire, Faustine Ngila speaks to Shirley Collie, founder and chief executive of BioInformatiCo, about how Leova works, the potential for AI to reduce the cost and complexity of clinical trial preparation, the importance of data quality and regulatory standards, and the opportunities and limitations facing African health technology companies seeking to compete in the global clinical research market.

Here is the interview excerpt:

1. Leova came second in the 2026 CDISC AI Innovation Challenge, competing against international innovators. What distinguished your platform, and what does this recognition reveal about Africa’s capacity to develop AI technologies for the global healthcare industry?

Most approaches to testing a clinical database are theoretical, they check specifications on paper. We tested inside the real thing. Leova enters synthetic patient scenarios directly into a live database, observes how each edit check actually behaves, and traces every result back to the protocol clause it came from, through to the CDISC submission standards. The output is evidence, not assumption, that the database does what the protocol says.

The recognition shows that Africa doesn’t just host clinical trials, it can shape how they’re run. We built Leova from first-hand experience delivering large, grant-funded regulated studies where budgets are tight and data quality can’t be compromised. Solving problems under those constraints produces innovation that the whole industry can use.

2 .You claim Leova can reduce database testing and validation from five days of specialist work to one. What measurable reductions in development time, costs and errors have you recorded across actual clinical research projects?

The largest measured gain so far is in testing and validation. On a Phase 2 HIV vaccine study about to enrol participants in Boston and Cape Town, Leova reduced database testing and validation from five days of specialist work to one.

Beyond testing, our internal benchmarks indicate significant reductions in study build effort. Form specification time has decreased from four hours to one, database programming and configuration from eight weeks to one, and custom report creation from four weeks to one. These are preliminary results from internal test runs, which we plan to validate on a live project when we build a first-in-human HIV vaccine study ahead of enrolment in Q2 2027.

3. Clinical trial databases can take 10 to 20 weeks to prepare. Which tasks can Leova reliably automate, which still require human expertise, and how do you prevent AI-generated errors from compromising clinical research?

Much of those weeks goes on repetitive, specification-heavy work, and that’s where Leova helps most. It reliably produces first drafts of data collection forms, data validation rules, quality control checks and reports, each traced back to the protocol clause it came from.

Human expertise remains essential for the judgement calls that require context and experience. This includes interpreting ambiguous or incomplete protocol language, making study-specific design decisions, and applying sponsor conventions. Every output is reviewed and formally signed off by an experienced data manager before going live.

Our safeguard goes beyond review, Leova is not only designed to generate a database, but also to support its verification and validation. Once a database is built, we run synthetic patient scenarios through the live system to test how each validation rule behaves and compare the results against the protocol requirements. Any discrepancies are flagged before go-live, with the results captured in an evidence pack that reviewers, sponsors and auditors can inspect. Where possible, outputs are deterministic, helping ensure that the same protocol produces the same result.

4. How have US health funding cuts affected BioInformatiCo and its clients, and can AI-driven efficiencies realistically offset research funding losses without compromising quality or jobs?

The US funding cuts have hit BioInformatiCo and many of our partners hard. The impact isn’t confined to Africa, many global institutions collaborate on research conducted here, so delays and cancellations ripple across the whole network e.g. Several studies we were supporting were paused, and we have responded by diversifying our funding base, including through European-funded work such as our EDCTP3 grant.

AI efficiencies cannot realistically offset these losses. Clinical trials routinely cost many millions of dollars, and most of that cost isn’t something software can remove. Nor should AI replace people. Trials depend on clearly defined roles, and accountability for each one rests with a qualified person, not a model. What AI can do is change how those people spend their time, less on repetitive build and testing work, more on judgement and oversight.

Where AI matters is in making every research dollar go further. A wave of AI-assisted drug discovery is producing more promising candidates than ever before, and each one must be tested in clinical trials to show whether it is safe and effective. If trials stay as expensive as they are today, many of those candidates will never be evaluated. Lowering the cost of running a high-quality trial is how we make sure good science isn’t lost for lack of funding.

5. How could platforms such as Leova help African researchers lead more clinical trials locally, generate evidence relevant to African populations and reduce dependence on external research infrastructure?

Our vision of Leova is to make it easy for African researchers to set up reliable, high quality databases from protocols, giving every user the know-how of an institutional data coordinating centre.

It comes down to choosing the right tool at the right cost. Where a tool helps us support research, we should use it. I don’t think about tools from a nationalistic perspective. But when tools are too expensive for our research context, we need home-grown solutions to fill the gap, and that’s what we’re trying to build.

The gap between what the Global North is investing in AI and trial technology, and what’s being built affordably for investigators, keeps getting wider. We don’t want to reach a point where we can’t trust investigator-led research because substandard systems and processes cast doubt on the integrity of the evidence. That’s why we believe a lot of exciting, disruptive innovation can come from focusing on this part of the market.

6. With plans to expand into the US, Europe and Australia, who are your primary customers, and how will BioInformatiCo compete against established clinical data management providers and larger AI developers?

Our primary customers are investigators running grant-funded research. Our competitors largely overlook this segment, so we are investing in building solutions designed for it. We believe we can win by offering these investigators world-class, high-quality service at a price that fits their budgets. The same model also benefits biotechs, which need their capital to go as far as possible.

7. Clinical trial data must meet stringent privacy, quality and regulatory requirements. How does Leova protect confidential patient information, maintain auditable records of AI-generated decisions and meet applicable regulatory standards, including US FDA requirements?

Leova is designed so that patient data never needs to pass through it. It works from the study protocol, which contains no patient information, and converts that protocol into the specifications needed to build a study database, including data collection forms, edit checks and database configuration. Patient data is collected and stored in the sponsor’s validated database system, not in Leova. Our closed-loop testing uses synthetic scenarios rather than real participant records, keeping confidential patient information entirely outside the AI workflow.

Every AI-generated output can be traced. Users review and approve outputs at each stage in the platform, and each build is tested against the protocol before a data manager signs it off for go-live. The test results and approvals are captured in an evidence pack, giving sponsors and auditors a clear record of what Leova produced, how it was tested and who approved it.

Leova adds to the regulatory controls sponsors already rely on rather than replacing them completely. The sponsor’s database system provides the 21 CFR Part 11 controls for electronic records and signatures, while Leova makes the database build faster and its quality demonstrable. This approach, with human oversight and documented testing throughout, fits ICH E6(R3) GCP and the FDA’s risk-based approach to using AI in drug development.

8. BioInformatiCo plans to open a seed funding round. How much capital are you seeking, what will the funds support, and what commercial milestones do you expect to achieve over the next 12 to 24 months?

We are seeking to raise US$2.3 million to further support the build-out of our platform and grow the number of design-partner customers working with us on it, as well as our paying customers, to set us up successfully for a Series A.

BioInformatiCo’s services revenue covers our day-to-day operations, allowing the seed round to be invested directly into Leova’s development and growth. Approximately 70% of the funding will go towards product engineering, 15% towards validation and strengthening our evidence base, and the remaining 15% will be held in reserve to provide around six months of runway ahead of a Series A raise.

Over the next 24 months, we aim to have three design-partner CROs on paid pilots and to publish a defect-detection benchmark within six months. We aim to reach five paying customers at 12 months and eight at 18 months, by which point customers will be running verification themselves. By month 24, we’re targeting ten paying customers, Leova in production on at least two database platforms, and our Series A underway.

Stay ahead of the Stories shaping our world. Subscribe to Impact Newswire and join our 
WhatsApp Channel for updates on global tech, business, and innovation—all in one place.

Dive deeper into the future with the Cause Effect 4.0 Podcast, where we explore the ideas, trends, and technologies driving the global AI conversation.

Got a story to share? Contact Us to reach a global audience with Impact Newswire.