Bucket A: Diagnostic Capability

A4 (integrated state representation) has the highest leverage relative to its risk, and is the answer for a funder seeking the most reliable return.

Three reasons. First, it has no fundamental scientific blockers — it is an engineering and data-infrastructure problem, classified low-risk on a 5-10 year horizon, so investment translates fairly directly into progress rather than betting on a breakthrough. Second, it is the substrate that makes every other diagnostic capability useful: without an integrated state representation, the outputs of A1, A2, A3, and A5 are heterogeneous data streams that must be interpreted manually. A4 is what turns four separate sensors into one clinical picture. Third, progress is happening anyway because of broader medical-AI development, so targeted investment compounds with industry momentum rather than having to carry the whole cost.

By contrast, A2 (whole-body molecular mapping) has higher potential impact but much higher risk, and accelerated funding does not reliably compress its timeline because the problem needs physics-level breakthroughs that are not yet visible. The advice that falls out of this: fund A4 as the high-confidence backbone; fund A2 as genuine long-horizon speculative research, understanding it may not yield within the programme. A3 is also low-risk and worth steady funding, but its leverage is narrower than A4's because it produces data rather than the integration that makes data actionable.

To a substantial degree, yes — and that independence is one of Bucket A's main strategic attractions.

Diagnostic capability has standalone clinical value regardless of whether AIHS-grade interpretation and therapy ever arrive. Better cancer detection, richer critical-care monitoring, earlier intervention in chronic disease — each of these improves outcomes on its own, with no dependency on the rest of the AIHS vision being realised. This is why the study repeatedly identifies Bucket A as the most defensible near-term investment even for a funder sceptical of full AIHS.

The independence is not total, however. A4's integration work depends on the data infrastructure that B5 provides, and shares the continuous-learning challenge discussed there. A5's engineered sentinel systems draw on the same synthetic-biology techniques as C1 and C5. So tight coordination across buckets accelerates all of them.

The realistic strategy is therefore: fund Bucket A for its standalone diagnostic value, treat cross-bucket coordination as a multiplier rather than a prerequisite, and recognise that the diagnostic advances will reach clinical usefulness on their own timeline whether or not the harder interpretation and therapeutic buckets keep pace. A diagnostic-only outcome is still a major medical advance.

Several capabilities a reader might expect to see are deliberately excluded, and the exclusions are principled rather than oversights.

Imaging hardware (faster whole-body MRI, next-generation PET scanners) is treated as infrastructure that AIHS would use, not as a capability the study tracks — the advances are about what must be newly achieved, not about the broad improvement of existing modalities. Population-scale epidemiological surveillance is a different problem entirely: it informs AIHS but is not part of any individual patient's closed diagnostic loop. Pre-conception and carrier screening, genetic counselling, and similar are important clinical genetics but are not closed-loop diagnostics for an existing patient in care.

The five advances in Bucket A are specifically the diagnostic capabilities an AIHS platform would need to operate as a closed loop — continuous, in-situ, integrated sensing of a single patient's state. They are not a complete picture of diagnostic medicine.

This scoping discipline matters for two reasons. For funders, it clarifies that Bucket A investment is complementary to, not a substitute for, broader diagnostic-medicine funding. For researchers, it keeps the study focused on the genuinely AIHS-specific gaps rather than diffusing into all of diagnostics. The four-parameter capability table at the top of this page (temporal resolution, spatial resolution, molecular breadth, integration) is the actual scoping test: an advance is in Bucket A if it is needed to move those parameters to AIHS-grade.

The bucket assessment notes that a system delivering roughly 70% of AIHS diagnostic requirements is achievable within 15-20 years even without solving the hardest advance. It is worth being concrete about what that partial system actually consists of, because it is the realistic deployment target rather than a consolation prize.

The achievable ~70% combines: mature continuous temporal sampling (A3, low-risk) giving sub-minute multi-parameter monitoring through implants and in-pod sensors; a strong integrated state representation (A4, low-risk) fusing those streams into a unified clinical picture; partial whole-body mapping (A2) via the composite approach described on its page — organ-resolution imaging plus single-cell-resolution sampling from accessible compartments plus cfDNA signatures — rather than true whole-body single-cell resolution; and distributed detection (A5) for the pathogen and damage classes that are tractable.

What the ~70% system lacks is the hardest 30%: true single-cell-resolution readout everywhere in the body in real time (the full A1 and A2 specifications), which needs the physics breakthroughs not currently visible.

Crucially, this partial system is not a failed AIHS — it is a transformative diagnostic platform in its own right, catching disease earlier and monitoring it more continuously than anything in current practice. Its value does not depend on the remaining 30% ever being solved. This is the concrete form of the study's recurring argument that Bucket A pays off on its own merits.