A new artificial intelligence (AI)-based framework developed by Indian researchers could help scientists identify elusive cancer stem-like cells that play a key role in tumour recurrence, metastasis and resistance to treatment. The technology may also support precision cancer research in regions where advanced single-cell sequencing facilities are limited.
Cancer treatment has made significant progress, but the disease remains difficult to control because a small population of tumour cells can survive even after therapy destroys millions of cancer cells. These surviving cells can contribute to tumour regrowth, spread to other organs and resistance to subsequent treatments.
Cancer stem-like cells are believed to be among the key drivers of these processes. However, detecting them has been challenging because they are rare and can continuously shift between different biological states.
Researchers at the S. N. Bose National Centre for Basic Sciences (SNBNCBS), an autonomous institute under the Department of Science and Technology (DST), Government of India, in collaboration with Ashoka University, have developed a new AI framework designed to address this challenge.
Led by Dr. Shubhasis Haldar, the research builds on the team’s earlier AI platform, OncoMark. The earlier system was developed to identify biological characteristics associated with cancer progression across millions of cells and achieved more than 99 per cent predictive accuracy.
Mapping Different States of Cancer Stem-Like Cells
The new platform, named ACSCeND (AI-based Cancer Stem-like Cell Profiler and Neoplasm Deconvoluter), moves beyond conventional approaches that generally provide a single measure of tumour stemness.
Instead, ACSCeND identifies three developmental states of cancer stem-like cells: pluripotent-like, multipotent-like and unipotent-like. This allows researchers to examine tumour heterogeneity in greater detail and understand how different stem-like populations may influence disease progression.
The framework combines biological information obtained from high-resolution single-cell sequencing with deep-learning techniques. Importantly, it can use this knowledge to analyse conventional bulk tumour RNA sequencing data.
This capability could significantly expand the number of patient samples that can be studied. While single-cell sequencing provides detailed information about individual cells, it is not always available because of cost, infrastructure and technical requirements. ACSCeND can therefore help researchers investigate hidden cancer cell populations in thousands of samples where single-cell data is unavailable.
Analysis of More Than 25,000 Tumour Samples
The researchers tested ACSCeND against existing computational approaches using independent datasets and different sequencing platforms. According to the study, the framework consistently performed better than current methods.
The team subsequently used ACSCeND to examine more than 25,000 tumour samples from major international cancer databases, including The Cancer Genome Atlas (TCGA) and PRECOG.
The analysis indicated that tumours containing higher levels of highly potent, pluripotent-like cancer stem cells were associated with poorer survival outcomes. Such tumours also showed a greater likelihood of recurrence and weaker responses to modern immunotherapy.
Beyond identifying stem-like cancer cells, the AI system helped reveal molecular programmes that may allow these cells to survive, adapt and evade immune responses.
Potential for Precision Medicine
The findings could have implications for future cancer treatment. Identifying patients whose tumours contain aggressive stem-like cell populations may help researchers assess relapse risks and understand why some cancers respond poorly to treatment.
The molecular pathways identified through the framework could also provide clues for developing new therapeutic targets and designing treatments tailored to the biological characteristics of individual tumours.
The research highlights the growing role of AI in biomedical science. By analysing enormous genomic datasets and detecting patterns that are difficult to identify manually, AI-based tools can help researchers better understand tumour biology and potentially accelerate the development of improved diagnostic and therapeutic strategies.
ACSCeND therefore represents another step toward using artificial intelligence to uncover hidden features of cancer and bring more precise, data-driven approaches to cancer research and care.
Also Read: CSIR-CRRI Showcases Indigenous MSS+ Technology for Sustainable Road Maintenance in Rajasthan
Author: Shivam
Shivam Dwivedi is a senior journalist with extensive experience in research-driven journalism, policy communication, and multi-platform storytelling. His areas of interest include international relations, defence, science & technology, education, urban development, agriculture, spirituality, and environmental sustainability. His work focuses on in-depth analysis, public discourse, and impactful narratives across governance and development sectors, with a strong commitment to the Sustainable Development Goals (SDGs). Contact: [email protected]
