'Trench AI' and Clinical Adoption: Closing the Healthcare Gap

For years, the digital health narrative in Mexico was built on monumental expectations: We hoped that the great decisive step would come from the hands of institutions. We imagined an ecosystem where Electronic Health Records (EHR) would finally be homogeneous and inter-institutional, where digital prescriptions would flow seamlessly between public and private sectors, and where robust telemedicine would finally manage to relieve the pressure on waiting rooms in our public hospitals.

However, history has taught us that systemic modernization is a challenge of very gradual changes. Infrastructure barriers, the complexity of regulatory frameworks, and interoperability challenges have shown that these changes require long-term adoption cycles. This background is vital to understanding the current moment of artificial intelligence: If we wait for AI to be adopted first as an institutional policy or a structural change in public operation, we run the risk of getting trapped in the same technological waiting room of the last decade.

But while the macro consolidates, a "pocket adoption" is emerging that doesn't ask for permission. This is "trench AI:" lightweight, high-precision solutions that are already in the hands of healthcare personnel. Doctors who don't wait for their hospital to update to use language models that synthesize scientific literature, or to employ computer vision tools that turn a smartphone into a diagnostic tool. Technological adoption is happening from the bottom up, empowering health professionals at the point of care, and solving clinical problems with the technology they already carry in their coat.

Lessons on Systemic Adoption

The history of health digitization in Mexico is, in many ways, a history of waiting. To gauge the challenge that artificial intelligence represents today, it is imperative to analyze the technological pillars we have tried to cement in recent decades. Fundamental projects such as the EHR, digital medical prescriptions, and telemedicine have been part of the national agenda for years, yet their consolidation remains partial and fragmented.

The case of the EHR is, perhaps, the most illustrative. Although the regulatory framework (such as NOM-024-SSA3-2012) has sought to regulate it for more than 14 years, real adoption in general clinical practice barely exceeds 40%. What is even more critical: interoperability, the ability for a patient's data to flow seamlessly between public or private institutions, remains a distant horizon. Today, the Mexican health system is an archipelago of digital silos that do not communicate with each other.

This implementation gap is not the result of a lack of vision or intention, but of a systemic complexity that is divided into three critical fronts:

• The regulatory time lag: The creation or modification of an Official Mexican Standard (NOM) can take 3 to 5 years. In a sector where technology evolves in cycles of a few months, regulation is often born obsolete. This creates a vacuum where powerful AI tools do not find a clear legal framework to operate, slowing down investment and institutional confidence.

• Opportunistic vs. structural adoption: As the FUNSALUD report in 2022 well observed, the COVID-19 pandemic forced a peak of adoption in telemedicine. However, it was an "opportunistic" adoption: an emergency lifeline that was not accompanied by process reengineering or administrative training. By not being a gradual and planned transition, many institutions returned to analog inertia once the urgency subsided.

• The resistance to the additional "click:" The health system is one of the country's largest employers, with a robust union and operational structure. If a digital tool is perceived as an additional bureaucratic burden or a threat to medical autonomy, adoption stalls. Without homogeneous training that demonstrates that technology reduces the administrative burden, new tools or software end up falling into disuse.

This panorama leaves us with a lesson: if we wait for the digital transformation of AI to be guided exclusively by structural changes, the health gap in Mexico will continue to grow. However, there is a parallel path that is already yielding results: decentralized adoption.

The 'Trench AI'

While the debate on systemic regulation continues, in the daily life of the consultation room, a pragmatic adoption that we could call "trench AI" has emerged. This is not an institutional imposition; it is an organic response from the health professional who seeks to recover their most valuable asset: time.

This small adoption is occurring under two fundamental premises that are digitally re-educating our sector:

• Lightening the operational and administrative burden. A doctor's first contact with AI is not usually a complex diagnostic tool, but solutions that eliminate operational friction. Today, health personnel are using algorithms to summarize scientific literature in seconds, transforming hours of updating and exhaustive bibliographic searches into protocols applicable to the patient in front of them. We are seeing the emergence of assistants that structure the medical note through voice, allowing the doctor to look the patient in the eye again instead of at the screen.

• The augmentation of clinical senses. AI is acting as a "multiplier of capabilities." It is no longer just about managing schedules, but about expanding what the human eye can detect. For example, through the use of computer vision applied to retinal images — an area where we have worked deeply — a first-contact doctor can now identify signs of chronic diseases that previously required high-specialty equipment. The same occurs with photographic analysis in dermatology or the automated calculation of drug interactions.

The most valuable aspect of this trend is not the technology itself, but the change in mindset. Every time a doctor validates an analysis generated by AI, a silent process of algorithmic literacy is occurring. The professional learns to distinguish where the tool is infallible and where their clinical judgment is irreplaceable.

This phenomenon is creating a health sector that is much more resilient and technified by conviction, not by decree. AI is not arriving to replace medical intuition, but to free it from tasks that cloud it due to saturation, returning the doctor to their most essential function: human care.

Clinical Democratization and Expert Judgment

The true potential of trench AI does not lie in the sophistication of its code, but in its capacity to decentralize high specialization. In a country where the distribution of specialists is profoundly unequal, technology acts as a bridge.

When we talk about bringing retinal image analysis to the first level of care, we are not only providing a tool; we are democratizing the capacity for early detection of chronic diseases that would otherwise reach the public system in advanced and costly stages.

However, for this democratization to be effective, we must address the challenges of trust and bias.

• AI as a “co-pilot,” not a “pilot:” Ethical adoption starts from an undeniable premise, “AI does not diagnose, the doctor does.” The tool offers an interpretation based on massive data patterns, but it is the professional who contextualizes that result within the patient's life history.

• Algorithmic transparency: The doctor does not need to know how to program, but they do need to understand the limits of the tool. Continuous training should no longer be about "how to use the software," but about how to interpret its signals and when to question them.

• Field validation: Trust is built with results. By integrating computer vision tools or language assistants into daily practice, the doctor validates the technology in the "real world," adjusting its use to the needs of the Mexican population, which has genetic and socioeconomic nuances that global models sometimes ignore.

Ultimately, the democratization of health through technology is an exercise in shared humility: the machine contributes greater speed and potentiates precision, while the doctor contributes ethics, empathy, and critical judgment.

The Health Team as the Architect of Trust

While the health ecosystem continues to work on regulatory frameworks that provide structure and certainty to the institutional use of Artificial Intelligence, this technology is already part of daily clinical practice. Decentralized adoption is not a risk to be curbed, but an opportunity that must be oriented.

True leadership in digital health today is not about waiting for the tool to be perfect, but about fostering a culture of responsibility where health professionals, with critical judgment and ethics, assume leadership of the clinical outcome