When Chang Gung Memorial Hospital in Linko, New Taipei, recently offered a monthly salary of NT$650,000 (approx. US$21,145) to recruit a pathologist, it sparked widespread debate about physician shortages. The case underscores a critical challenge in healthcare: staffing gaps are not confined to nursing but extend to highly specialized physician roles.
Samuel Chen, co-founder and CEO of AIxMed, noted that the shortage is a global problem. Over the past decade, the number of practicing pathologists in the US has declined by more than 20%, leaving the remaining workforce to shoulder heavier workloads. With an aging cohort of specialists and limited new entrants, generational transition is proving difficult.
Unlike cardiologists or neurologists, pathologists rarely interact directly with patients, working largely "behind the scenes" to confirm diagnoses, particularly in oncology. Yet their role is indispensable in cancer medicine, where shortages are already placing pressure on clinical care.
AI boosts accuracy but does not replace physicians
Chen highlighted findings from a 2016 U.S. study showing a 3.5% error rate for manual interpretations by pathologists, compared with 2.9% for software. Crucially, when physicians used AI tools, the error rate dropped to just 0.5%. He argued that this is evidence that human-machine collaboration offers the best solution.
In 2018, Chen and Dr. Tien-jen Liu of MacKay Memorial Hospital co-founded AIxMed to harness AI for cytological slide analysis. Their applications include detecting bladder cancer through urine cytology and thyroid cancer via fine needle aspiration biopsies, where AI can identify malignant cells with higher sensitivity than human visual inspection alone.
"AI in medicine is an assistant, not a replacement," Chen emphasized. "Clinical judgment must remain in human hands."
From tech industry to medical innovation
Before founding AIxMed, Chen built a career across global tech firms, including Sun Microsystems, Trend Micro, Asus, and Compal Electronics. At Compal, he launched the company's AI lab in 2015 and later entered the medical field after earning a silver medal among 4,000 teams in a Kaggle data science competition focused on cell image analysis.
Encouraged by mentors to pursue medical AI outside of the electronics sector, Chen partnered with Dr. Liu, merging their expertise in technology and pathology. The company name, AIxMed, reflects the fusion of their expertise.
Since then, AIxMed has targeted the US market as its primary focus while working with Taiwanese hospitals on clinical validation. The company has also begun exploring partnerships with Japanese scanner manufacturers to integrate hardware and software solutions.
Domain knowledge sets competitive edge
While AI tools such as those from Nvidia are commercially available, Chen stressed that successful medical applications require extensive integration and training with hospital partners.
Asked about Alphabet's announcement that its Med-PaLM model had passed the US Medical Licensing Examination, Chen expressed doubts about its practical value, suggesting the AI may be relying on memorization rather than true reasoning. Instead, AIxMed is building proprietary models evaluated by top medical experts, including cellular pathology authority Barbara Crothers, who serves as the company's chief scientist.
Next step: From image analysis to AI agents
Looking ahead, Chen envisions moving beyond image recognition toward AI agents that can combine reasoning abilities with pathological expertise. Such agents could support family history risk assessments, recommend testing protocols, and generate comprehensive diagnostic reports with the help of generative AI, further strengthening the role of AI as a trusted assistant to pathologists.
Article edited by Jack Wu