What the total combines
What interests me in Mexico’s Social Security Institute announcement is what increased imaging activity changes for a patient. IMSS disclosed more than 27.6 million examinations in 2025. Its specialist Víctor Mora Santiago attributes productivity gains exceeding 50%, and reaching 150% in some processes, to AI together with new technology. That is a useful starting point for understanding the workload of a hospital service. The measurement still needs a clear meaning: counting completed procedures does not by itself tell us how much earlier a patient receives an actionable result.[1]
The total combines 19.8 million X-rays with 498,000 magnetic resonance imaging (MRI) examinations. That mix makes the denominator in any productivity calculation consequential. Extending a gain in one process to an entire service assumes that different examinations can meaningfully be counted on the same basis. I would first ask where the improvement was measured: image acquisition, report preparation or completed daily work? Applying the 150% figure for certain processes to the whole radiology service would go beyond the announcement. Results broken down by examination type would reveal differences concealed by the aggregate.[1]
Two plausible explanations
One explanation is that AI has eased workflow bottlenecks. Another is that new equipment and service organisation account for more of the improvement. In the same announcement, IMSS describes CT systems with 256 to 512 detectors and 3-tesla MRI equipment. An evaluation seeking to isolate AI would have to account for these concurrent changes. The workforce of 5,885 radiology technicians and 2,634 radiologists also underlines the human labour involved. The aggregate figures do not distinguish which of these plausible mechanisms contributes most.[1]
To compare those explanations, I would favour units performing the same type of examination with comparable patient groups and equipment. The difference between workflows with and without AI would then be more informative. This is a proposed comparison, not a claim that IMSS has conducted such a study. Its start and finish points would also need to be fixed: time from referral to a signed report answers a different question from scanner acquisition time. A design that also tracks workload and staffing could show the conditions under which a productivity gain recurs.[1]
When the report reaches the patient
I would not discount the value of reducing a patient’s wait. Extra capacity can itself be a worthwhile objective in a busy imaging service. A claim about diagnostic precision nevertheless requires additional outcomes. My proposed evaluation would include repeat scans, false positives, false negatives and missed urgent findings. These are not presented as measured outcomes in the announcement. Knowing how they change alongside faster reporting would help establish what the gain means for patients. Each should be interpreted with the examination type and clinical need in view.[1]
The IMSS announcement offers a starting picture of a large service changing alongside equipment upgrades and AI. The next result that would raise my confidence would show shorter time to a report for a defined examination group, together with how errors changed in that same comparison. Replication across other units would support moving from an aggregate capacity claim to a benefit for a particular clinical workflow. Until then, keeping the productivity claim within its stated scope takes service improvements seriously while leaving the size of any diagnostic benefit open.[1]