Cancer data isn't diverse enough, and it is costing lives

The lack of diversity in data is one of medicine’s most persistent blind spots, and especially so in cancer research. PAICON, a healthtech company born at the intersection of artificial intelligence and oncology, is tackling precisely that problem, ensuring that AI-powered diagnostics work not just for the few, but for the ‘84 per cent’ of the world’s population that current models overlook.  

“Tumour biology, genetics, and treatment response vary widely across populations, yet most AI models are trained on data representing only about 16 per cent of the world,” says Dr  Manasi A-Ratnaparkhe, CEO and co-founder of PAICON. “That leaves the remaining 84 per cent underserved and often misdiagnosed. Our mission is to change that.” 

PAICON showcased its AI-powered oncology solutions at the debut edition of WHX Tech in Dubai and participated in its start-up competition Xcelerate, which featured 50 startups presenting innovations and competing for a cash prize of $50,000.

A global cancer data lake and a fairer future 

At the heart of PAICON’s proposition lies the PaiX Cancer Data Lake, a curated, harmonised collection of more than 130,000 cancer cases spanning over 60 countries. In Ratnaparkhe’s words, “It is the world’s first truly global cancer data lake for AI diagnostics.” 

Related:Gulf set for ‘quantum leap’ in AI-driven drug discovery

Diversity determines accuracy. Explains Ratnaparkhe, “Algorithms trained on narrow, Western-centric datasets fail to recognise genetic variants, tissue morphologies, or biomarkers that are prevalent in other regions and that is a flaw that can cost lives.” 

The case of Dr Anil Kapoor, who died after an adverse reaction to chemotherapy due to an unrecognised genetic variant absent from standard Caucasian-centric tests, underscores the urgency. “It’s not just a data gap, it’s a matter of life and death,” Ratnaparkhe stresses. 

Accelerating diagnosis from weeks to hours 

PAICON’s technology plugs directly into the clinical pathway, transforming how oncologists and pathologists work. Its flagship model, SatSight DX, can detect microsatellite instability in colorectal cancer straight from a digital slide, reducing diagnostic turnaround from three weeks to about an hour, and cutting costs from roughly €500 to less than €10. 

“What once required costly molecular sequencing can now be done in about an hour,” Ratnaparkhe notes. “It’s precision oncology made practical.” 

Another innovation, PaiNet, connects oncologists worldwide for AI-assisted second opinions. It fuses real-time AI slide analysis with human expertise, creating a hybrid system that combines machine precision with clinical intuition. 

Related:Pharma can help build health systems, not just sell products

Regulatory rigour and ethical oversight 

Healthcare may be global, but regulation continues to remain local. PAICON’s strategy reflects that reality. Its AI products are developed under an ISO 13485-certified quality-management system, ensuring traceability and reproducibility from concept to deployment. 

Every model undergoes multi-site clinical validation in collaboration with hospitals and cancer centres worldwide, with strict adherence to GDPR, HIPAA, and the EU AI Act. For countries that require data localisation, PAICON employs federated or on-premise architectures, allowing algorithms to learn without moving sensitive patient data. 

Such design aligns with emerging global frameworks like the WHO’s guidance on ethics and governance of AI in health (WHO, 2021), which calls for fairness, transparency, and accountability across the AI lifecycle. 

Building partnerships across the healthcare spectrum 

Early adoption has been strongest among oncologists and pathologists in Europe, India, and the Middle East. Hospitals report faster access to molecular insights and improved diagnostic confidence. 

The enthusiasm is not limited to clinicians. Pharmaceutical firms and public-sector agencies are eyeing PAICON’s diverse data as a means to improve drug repurposing, biomarker discovery, and clinical-trial diversity, which have been some of the long-standing weak points in the biopharma pipeline. 

Related:Data-Driven Innovations: Shaping the Future of Human Health and Longevity

“Our roadmap rests on collaboration,” says Ratnaparkhe. “Hospitals bring clinical context, pharma brings research scale, and public institutions bring data stewardship. Together, we can make AI adoption both ethical and effective.” 

The model mirrors initiatives such as the AI Lab of UK’s NHS and the European Health Data Space, which promote shared infrastructure and standards for responsible data use. 

The economics of equity 

Beyond the moral case, PAICON’s approach makes economic sense. By slashing diagnostic time and cost, tools like SatSight DX not only improve outcomes but also free up resources. “On a system level,” Ratnaparkhe explains, “It translates into shorter patient wait times, fewer unnecessary procedures, and lower treatment expenditure.” 

The ripple effects are especially significant for low- and middle-income countries, where sequencing infrastructure is limited and healthcare budgets are tight. Democratising access to molecular-level precision could redefine cancer care economics. 

A responsible-by-design perspective 

Amid global debates on AI ethics, bias, and transparency, PAICON positions itself as a standard-bearer for responsible AI. “Bias begins where representation ends,” says Ratnaparkhe. The company’s diverse datasets, human-in-the-loop validation, and traceable model development are designed to mitigate that risk. 

Platforms like PaiNet combine algorithmic predictions with expert oversight, ensuring interpretability and accountability. “Responsible AI at PAICON means designing technology that doctors can trust, patients can rely on, and regulators can verify,” she adds. 

The five-year vision 

By 2030, PAICON wants to redefine the benchmark for ‘global standard’ in cancer AI, one that is both technologically advanced and globally representative. 

“If a patient in Santiago, Dubai, Nairobi, Mumbai, Tokyo, or Munich receives the same diagnostic accuracy from our AI, that will be our true milestone,” Ratnaparkhe says. 

She envisions the PaiX Data Lake powering a whole ecosystem of AI models, including PAICON’s own and those of partners, while influencing regulatory frameworks to make data diversity a clinical expectation, not an afterthought. 

The company’s message resonates at a pivotal moment as investment for AI in healthcare are witnessing a meteoric rise with every year. Global digital health funding has reached $20.8 billion in the first nine months of 2025, according to Galen Growth, the global authority in Digital Health intelligence, and yet concerns around bias and generalisability persist. 

As medicine enters an era where diagnostics depend as much on datasets as on microscopes, PAICON’s work underscores the importance of data. If data isn’t representative of the world, the AI won’t be either.  

By building technology that reflects humanity’s full genetic and geographic diversity, PAICON is trying to improve algorithms and truly expand the reach and efficacy of modern healthcare.