Researchers at UC San Diego published CANDiT, a machine‑learning systems‑biology framework that identifies differentiation therapy targets to reprogram cancer stem cells (CSCs) and trigger self‑destruction in colorectal cancer models. The tool maps gene networks from key lineage genes and nominates druggable nodes to restore differentiation markers like CDX2. Complementary reporting described precision reprogramming strategies that manipulate tumor plasticity to halve recurrence risk in colon cancer models. Both efforts use computational approaches to pinpoint vulnerabilities in CSCs—a population linked to metastasis and therapy resistance—and illustrate a push to translate AI‑driven target discovery into differentiation‑based therapeutic strategies.