A newsletter by SPJIMR WISE Tech

Volume 1 | Edition 5 | February 22, 2025

Quest examines innovation, technology adoption, and societal impact through a multidimensional Ethics-Responsibility-Sustainability (ERS) lens. A contemporary innovation-related event or trend in society triggers each narrative, be it an article, a case study, or an interview story. In response, the narrative raises a thought-provoking ERS question on the consequences of that event or trend in the future and initiates a discussion on what that may imply for managers and policymakers across a spectrum of businesses—from start-ups to large corporates.

AI in cancer detection:
Promise of a revolution?

Article

Meredith, a 40-something data journalism professor at the Arthur L. Carter Journalism Institute at New York University, became one of the first beneficiaries of AI-assisted diagnostics for breast cancer detection. Her routine mammogram initially showed inconclusive results, but an AI system flagged subtle abnormalities that radiologists had missed. Further testing confirmed early-stage breast cancer, and it was successfully treated. As Meredith reflects, the AI system likely saved her life by detecting cancer at a stage where it was most treatable¹.

As cancer becomes one of the most significant health challenges globally, these ideas require urgent exploration to ensure that AI-driven innovations serve all communities, regardless of socio-economic disparities.

This real-life case underscores the transformative potential of AI in improving early cancer detection, especially in cases where traditional diagnostic methods might fall short.

Meredith Broussard, a researcher and advocate for data analysis for social good, used her experience to write More Than a Glitch², a book that critically examines AI’s role in society. She explores fundamental ideas of equality, equity, and responsibility embedded in AI, particularly in sensitive domains like healthcare.

Power of AI in cancer detection

Early detection is a cornerstone in fighting cancer. Studies indicate that when cancer is detected early, treatment can begin promptly, often leading to less invasive therapies and improved patient outcomes³. In fact, every month that cancer goes undetected and untreated, a patient’s chances of survival decrease by 10%⁴.

Early diagnosis leads to a greater probability of survival with less morbidity, as well as less expensive treatment⁵. – WHO

AI-driven diagnostics are proving invaluable in this area, assisting in the detection of not only breast cancer but also lung, brain, prostate, skin, and thyroid cancers. Advanced AI-powered imaging tools, such as machine learning algorithms for CT scans and MRI analyses, are significantly improving early detection across various cancer types.

Trained on data from thousands of images and sometimes supported with information from a patient’s medical record, AI tools can tap into a larger database of knowledge than any human can. AI can scan deeper into an image and pick up on properties and nuances among cells that the human eye cannot detect (like in Meredith’s case). When it comes time to highlight a lesion, AI uses different colors to point out different levels of abnormalities such as extreme cell density, tissue calcification, and shape distortions.

AI-driven cancer detection is transforming oncology by enhancing diagnostic accuracy and accessibility. By analysing vast datasets, AI assists in early identification, reducing misdiagnoses and bridging healthcare gaps worldwide.

“AI tools help us assess what course of treatment might be most effective, based on the characteristics of the cancer and data from the patient’s medical history”, says Tufia C. Haddad, MD, medical director of digital strategy at the Mayo Clinic Center for Digital Health in Minnesota, whose radiology and pathology colleagues at the clinic use AI for cancer diagnostics⁷.

When the diagnosis is negative for cancer, AI tools can help avoid unnecessary follow-up biopsies. AI-assisted analyses helps reduce false positive diagnoses⁸.

Wish to read later? Receive Quest in your inbox. >

Future of AI in cancer detection

AI is emerging as a transformative tool, leveraging large datasets to enhance diagnostic accuracy and efficiency. According to a report by Grand View Research, the global market for AI in cancer detection is expected to grow at a compound annual growth rate (CAGR) of 32.4% from 2022 to 2028⁹.

Concerns about the accessibility, affordability, and inclusivity of these technologies need to be addressed, especially from the perspective of emerging economies¹³.

This growth is fuelled by several factors, such as including:

a) The increasing accuracy and efficiency of AI-powered cancer detection tools

b) The decreasing cost of AI technology

c) The growing awareness of the benefits of early cancer detection

d) The increasing investment in AI R&D.

However, global disparities are visible in healthcare access because survival rates vary drastically by region. For instance, in high-income countries, early-stage breast cancer has a five-year survival rate of nearly 99%, compared to only 27% for late-stage diagnoses¹⁰. In contrast, in countries like India, where early detection remains a significant challenge due to limited screening access and healthcare inequities, survival rates in breast cancer patients are far lower¹¹. Similar gaps exist for lung and prostate cancer detection, with AI-driven tools still being concentrated in technologically advanced healthcare systems¹².

Concerns around the use of AI in cancer detection

There are concerns about data inclusivity when it comes to how AI models are built and used in cancer detection. Most of these systems are trained on data from wealthier countries, meaning they don’t always account for genetic differences, cultural factors, or economic realities in emerging economies. This creates a risk that AI may work well for some populations but not for others, leading to unfair outcomes. Model bias—particularly related to demographic characteristics such as ethnicity, socioeconomic background, and genetic variations—affects the accuracy and reliability of AI-driven predictions in emerging economies. To make AI more effective in diverse healthcare settings, there is a pressing need to improve data representation by incorporating region-specific medical records and ensuring inclusive, bias-free training datasets.

The lack of transparency in how AI models make decisions is often called the black box problem.

Another concern is the lack of transparency in how these models make decisions, often called the black box problem. If doctors and patients don’t fully understand how AI reaches its conclusions, it becomes harder to trust the results, correct mistakes, or hold anyone accountable when things go south. One such case is the controversy surrounding IBM Watson for Oncology, where hospitals claimed the system provided incorrect treatment recommendations, leading to patient safety concerns¹⁴. Another example involves situations inviting lawsuits against radiology AI developers if alleged misdiagnoses might lead to delayed cancer treatments¹⁵. Clear accountability frameworks are essential to build trust and drive the adoption of AI in healthcare, and they are a hygiene factor in oncology treatments and cancer detection.

AI-driven cancer detection is advancing rapidly in urban hospitals, where strong infrastructure and specialised personnel enable easier adoption¹⁶. However, accessibility remains a major challenge—over 50% of India’s oncologists and cancer treatment facilities are concentrated in urban areas, leaving rural populations underserved¹⁷. A significant disparity exists in the distribution of healthcare professionals between urban and rural areas. Approximately 74% of doctors practice in urban regions, which are home to only 28% of the population, leaving rural areas underserved¹⁸. One way to bridge this gap could be integrating AI-based diagnostics with telemedicine, allowing remote consultations and screening camps in rural areas to reach patients in low-resource settings.

Come back later? Receive Quest in your inbox. >

Barriers to adoption of AI in cancer detection

Economic barriers: Advanced AI systems often come with high costs, raising concerns about whether they widen the healthcare gap. For instance, while the UK Department of Health and Social Care has invested £123 million into 86 AI technologies for healthcare delivery, many low-resource settings lack the funding for similar advancements¹⁹. Moreover, although AI tools like mammogram analysis demonstrate 99% accuracy and 30 times faster results, the upfront costs remain prohibitive for under-resourced healthcare systems²⁰.

Energy costs: Training and deploying AI models demand significant computational resources, contributing to environmental challenges. For instance, training a single large AI model can emit as much CO2 as five internal combustion vehicles over their entire lifetimes²¹!

Cancer detection through AI in India and abroad

Governments and the private sector are making significant strides in leveraging AI for cancer detection and diagnosis. Start-ups like Qure.ai and Vara.ai are leading the charge by developing cost-effective and accessible diagnostic tools tailored for emerging markets. Google Health’s AI model for breast cancer detection has demonstrated groundbreaking accuracy, outperforming radiologists in identifying early-stage cancers. PathAI collaborates with global healthcare systems to improve pathology diagnostics using AI, while Zebra Medical Vision continues to innovate in AI-powered imaging for disease detection. IBM Watson Health has focused on oncology decision-support systems, enabling more informed treatment planning.

In India too, AI is playing a transformative role in cancer detection and treatment, with several notable initiatives:

These reflect India’s commitment to integrating AI into healthcare to enhance early cancer detection and treatment, addressing challenges such as limited access to screening and disparities in healthcare infrastructure.

At SPJIMR WISE Tech, we focus on wise innovation &
technology. And we define ‘wise innovation’ as purposeful
innovation done for the right reasons in the right way.
So, given the potential of AI, how can we enhance wise
adoption of AI in cancer detection?

Author of this edition: Siddharth Jain | | Researcher at SPJIMR WISE Tech

Sources

Email us at wisetech@spjimr.org. >

Coming soon…

Eyes in the sky:
Navigating the ethical
and societal impact
of drone technology

Upcoming editions of Quest

Trading emissions:
Unpacking
carbon credits

Read our earlier editions

Are nutra supplements
supplementing your
health?

What role will hybrids
play in the future as India
gears towards winder EY
adoption?

Fast Fashion: Fast
on fashion, slow
on responsibility?