Clinical diagnosis applied to STEM learning.
The shortest path between the problem and genuine understanding, with the fewest unnecessary detours.
In an ordinary classroom, the bus stops near home — but everyone gets off at the same place.
The lesson plan is built for the group. The route never changes to accommodate individual reasoning. The hard work of figuring out what is missing falls entirely on the student — alone, in the bus seat, with the teacher up front, driving toward a different destination.
What is missing is not the content. It is the diagnosis of how each student actually thinks.
An error is not a sign of incompetence. It is a technical signal — it carries information about which reasoning step failed, at which point in the cognitive structure, with which pattern of recurrence. When that signal is read with rigor, the teaching that follows does not have to guess. It operates on data.
That is what diagnostic engineering delivers: an operating system for the continuous reading of a student’s reasoning — correct answers included — showing which paths are efficient, which carry hidden misconceptions, which need consolidation, and which can be advanced.
Diagnostic engineering is not new. It is new applied to cognition.
The method carries a 24-year development trajectory. It was born in industrial engineering, adapted to hospital systems in peer-reviewed research, and since 2014 it has been applied to STEM learning. Each stage added rigor that survives into the next.
Six traceable steps, from error to recommendation.
The FMEA protocol applied to cognition operates in a fixed sequence. Each step produces output for the next. Every output is auditable. The analysis has a verifiable beginning, middle, and end — it is not subjective interpretation, it is instrumentation.
Failure mode identification
Each error is categorized into one of the dictionary’s 12 families (40 subtypes). The classification answers: what is the cognitive type of the failure, not merely the topic where it appeared.
Root-cause analysis
Each mode is decomposed down to the operational root. It is not enough to know the error is one of “partial reading” — one must know which missing protocol produced the partial reading.
Consolidation Index per Skill (ICS)
Quantifies the degree of consolidation of each skill — fluent correct answers, hesitant correct answers, lucky guesses, errors. It ranges from 0 to 1. ICS < 0.5 indicates insufficient consolidation; ICS > 0.85 indicates operational mastery.
Resolution Speed Index (IVR)
Measures temporal efficiency. IVR < 1.0 indicates fluent resolution; IVR > 1.5 indicates laborious reasoning or an inadequate strategy. Combined with right/wrong, it separates “knows it but is slow” from “does not know it and guessed”.
Risk Priority Number (RPN)
RPN combines the severity of the mode, the observed frequency, and the current detectability. It allows failure modes to be ranked by intervention priority — not everything is treated equally; the highest RPN becomes a dedicated lesson.
Operational recommendation
Each mode with a critical RPN generates a concrete pedagogical action: a dedicated lesson, question refraction, a verification protocol, a calibration exercise. The recommendation is specific, not generic.
Technical vocabulary
Student X — a synthetic profile built from real patterns.
A methodological composite of patterns observed across multiple students. No individual data is identifiable. The presentation follows the structure of the real reports delivered to families: first the reading of who the student is — in five movements — then the technical evidence that supports each move of the reading.
Profile
- Age
- 16 years
- Context
- International school, bilingual curriculum
- Initial diagnostic
- Full SAT practice test (Bluebook)
- Plan window
- 5 months, ~20 weeks
Score
Initial composition: Math 510 · R/W 540. Realistic projection discussed below.
Student X in five movements
Before the numbers, the closest reading of Student X as a person, in five movements — mirror, surprise, mechanism, evidence, horizon.
Failure modes identified — the evidence behind the reading
The narrative reading above is not subjective interpretation. Each move described maps to a quantified failure mode, ranked by priority.
| Mode | Operational description | ICS | IVR | RPN |
|---|---|---|---|---|
Dissociative mode Right setup, wrong execution | Reads the prompt, fires the correct procedure, but loses the link between the data read and the calculation executed. The error appears in execution, not in comprehension. | 0.34 | 1.42 | 428 |
Consolidation inversion Advanced mastered, fundamentals gapped | Apparent mastery in advanced topics, gaps in fundamentals. The classic pattern of an international curriculum that skips consolidation. | 0.41 | 1.18 | 385 |
Textual overload Drop on dense prompts | Performance drop proportional to prompt length. It is not a lack of vocabulary — it is the absence of a segmented-reading protocol. | 0.52 | 1.68 | 312 |
Self-imposed restriction Fixation on the first strategy | The student fixes on an initial strategy and does not test alternatives, even when the result contradicts it. Arbitrarily narrows the scope of reasoning. | 0.48 | 2.04 | 298 |
Pacing collapse Drop in the last 10 minutes | The accuracy curve drops abruptly in the final third of the test. It is not fatigue — it is the absence of prior pacing and a skip strategy. | 0.63 | 0.92 | 186 |
Domain Map — TARC calculator by area
Each cell represents an assessed sub-area. The color indicates the degree of operational consolidation. Red areas are absolute intervention priorities; green areas are the student’s cognitive assets and must be preserved.
Operational recommendation — derived from the critical RPNs
- Data-to-result verification protocol. For the dissociative mode (RPN 428), introduce a mandatory coherence check between the data read and the calculation executed, before marking an answer. Applied in the first 4 Math sessions, it becomes automatic by week 6.
- Consolidate fundamentals before advanced topics. For consolidation inversion (RPN 385), suspend exercises on topics where ICS is low and return to Heart of Algebra and Inference — areas with critical TARC in the domain map. Without consolidation here, advancing only amplifies the fragility.
- Segmented reading with anchors. For textual overload (RPN 312), introduce a reading protocol in 2-sentence blocks with re-reading anchors. Applied in Reading exercises from the first week of R/W.
- Strategic divergence training. For the self-imposed restriction (RPN 298), exercises with the explicit instruction “solve it two ways”. Breaks the fixation on the first strategy chosen.
- Pacing and skip protocol. For pacing collapse (RPN 186), strategic-skip training on full practice tests from week 5 of the Sprint.
What Student X recovers from a careful reading of this report alone
Even before the first conceptual remediation session, three takeaways that follow directly from this reading produce measurable, auditable gains:
- Recognition of the five trap families — the student begins to identify the shape of the distractor before marking. The effect is not uniform (it depends on triggering the protocol under time pressure), but for 2 to 3 additional Math questions per test, explicit familiarity is enough.+30 to +40 pts · Math
- Data-to-result verification protocol — between an intermediate step and the final answer, the student interrupts the silent automatism and checks coherence. It directly reverses the dissociative mode in cases where the calculation was correct and only the last step flipped.+20 to +30 pts · Math
- Evidence-versus-intuition discrimination discipline — in Reading, the student begins to ask “is this option supported by the text, or does it just resonate with my anticipation?” before marking. It directly neutralizes the self-imposed restriction of his repertoire.+30 to +40 pts · R/W
Realistic projection — with a plan tailored over the RPNs
Based on the typical absorption curve observed when the intervention is ranked by RPN, with a diagnostic review every 4 weeks. It starts from the score already internalized via the report (previous item), not from 1050.
The diagnostic is not a loose analysis. It is the input that defines every lesson that follows.
Each of Student X’s critical RPNs demands a specific lesson, with tailored material. It is not a selection of pre-existing exercises — it is production. The same failure mode in another student generates a different lesson, because the root cause is different. Below, an example of a lesson built over the dissociative mode in an IB Biology context.
IB MYP 4 Biology — Criterion C: Processing & Evaluating
Suggested: 3-4 screenshots of real lessons from the IB Criterion C Biology course,
or a 60-90 second video walking through a complete lesson.
The lesson presents the classic catalase-versus-temperature experiment, but with explicit pedagogical instrumentation: mandatory verification points between the data collected and the conclusion drawn. The student cannot advance without closing the coherence checkpoint. The design is diagnostic — the lesson itself works against the dissociative mode.
If your family, school, or institution needs diagnostic engineering for STEM learning — let’s talk.
The first step is a 30-minute demo session, at no cost. I present the method applied to the specific profile of the student or context, show the Domain Map live, and answer questions. It is not a sales pitch — it is a technical reading of what makes sense for your case.
Available in person in Leblon, or by international video call.