Decision Intelligence Preview · RWE Correlation Model
PRE-LAUNCH STRATEGY PREVIEW
Strategic Objective: Identifying "Latent Markets" where high Graves' Disease medication volume (NDB Proxy) suggests an untapped X Brand patient pool.
Click any prefecture on the map to view the local intervention window.
47
All prefectures
showing all
—
High priority
—
Medium priority
—
Divest
—
Confirmed burden
sub-segment
Show:
High priority
Medium
Divest
Confirmed burden
Latent
Surgical-heavy
Prefecture ranking
Sorted by composite opportunity score
Back to map
Click any prefecture on the map or in the list to see its indicator breakdown and recommended actions.
Source: NDB/MHLW 8th–10th edition · PMC literature review · X Brand launch Sep 2024Indicators: 23 · Outcome proxy: IVMP + orbital surgery rates per 100k · Pearson correlation
CONFIDENTIAL: Intended for the Client X Brand Launch Team. Based on Signal 0 (PMC Literature) overlays on MHLW NDB Open Data. Do not distribute externally.
X BRAND JAPAN: MARKET ILLUMINATION
Decision Intelligence Preview · RWE Correlation Model
PRE-LAUNCH STRATEGY PREVIEW
Strategic Objective: Identifying "Latent Markets" where high Graves' Disease medication volume (NDB Proxy) suggests an untapped X Brand patient pool.
Click any prefecture on the map to view the local intervention window.
X Brand Japan · Commercial Intelligence
Dashboard Quickstart Guide
How to read, navigate, and act on the prefecture opportunity dashboard — for commercial and sales leadership.
The dashboard is designed to be read left-to-right: the map gives orientation, the ranking list gives precise ordering, and the detail panel gives the evidence and action. You can drive it from any of the three panels — clicking a prefecture in the list updates the map highlight and the detail panel simultaneously.
02
Colour legend
Each prefecture is coloured by its opportunity segment — a combination of priority tier (High / Medium / Divest) and sub-segment type that determines the commercial action. Within High Priority, three sub-segments use distinct shades of green and blue to indicate which type of opportunity exists.
Confirmed burden
High TED-adjacent diagnoses combined with high IVMP volumes. Active moderate-to-severe TED patients are already being managed — but with the older standard of care. The most actionable switch opportunity for X Brand.
High Graves dxHigh IVMPHigh TRAb testing
Latent
High upstream thyroid disease indicators but low treatment signal. TED burden likely exists but patients are not reaching active management — possible diagnosis gap, referral gap, or specialist access barrier.
High Graves dxLow IVMPLow surgery
Surgical-heavy
High orbital decompression and strabismus surgery volumes alongside low IVMP. Patients are arriving at surgery having bypassed earlier medical intervention. X Brand's upstream prevention story is most compelling here.
High orbital surgeryLow IVMPLate-stage TED
Medium priority
Mixed or moderate signals across indicator tiers. Not a clear action now, but worth monitoring. A local KOL conversation can clarify whether the ambiguity reflects a true thin market or a data limitation.
Mixed signalsScore 35–65
Divest
Low signal across all indicator categories — diagnoses, treatments, and surgical procedures. The TED patient population appears genuinely thin in this prefecture. Resources are better deployed elsewhere in the current cycle.
Low all indicatorsScore < 35Low specialist density
03
Indicator tiers
Not all indicators carry equal weight. They are organised into four evidence tiers based on how directly each signal predicts TED patient burden, as established through the PMC literature review. Tier D indicators — treatment procedures — carry the highest analytic weight because they confirm active disease management, not just risk.
A
Upstream thyroid disease
Graves disease, Hashimoto thyroiditis, TRAb testing, antithyroid drugs, RAI therapy, thyroidectomy. Up to 40% of Graves patients develop TED. High A-tier volume means a large at-risk pool in this prefecture.
Weight: 0.8 · ICD E05.x, E06.3 · drug codes
B
Ocular symptom signals
Proptosis, diplopia, eyelid retraction, periorbital oedema, dry eye, dysthyroid optic neuropathy. These are clinical presentations of TED recorded as separate diagnoses — a strong indication patients are symptomatic but may not have a TED code.
IVMP high-dose infusion, orbital decompression surgery, strabismus surgery, eyelid surgery, orbital radiotherapy. These are the outcome variables — the model correlates all A/B tier indicators against these. High C-tier = confirmed active or advanced TED being treated in this prefecture.
Weight: 0.9 · K0681x surgery codes · methylprednisolone drug codes
D
Care pathway & specialist access
Endocrinology visits, ophthalmology visits, and critically — patients appearing in both specialties (co-visit signal). The co-visit is the highest-specificity care pathway indicator, as it mirrors the multidisciplinary management TED requires.
Weight: 0.4 · specialty codes 10 (内分泌) + 13 (眼科)
04
How the opportunity score works
Each prefecture receives a composite opportunity score from 0–100 built from three components:
Score = Σ
(normalised indicator rate per 100k)
×
|Pearson r with outcome|
×
tier weight
Normalised rate
Each prefecture's raw count is divided by population (per 100,000) and min-max normalised across all 47 prefectures. Prevents large prefectures like Tokyo from dominating solely due to size.
Pearson correlation
Each indicator's correlation with the outcome variables (IVMP rate, orbital surgery rate) across all 47 prefectures. Only indicators with statistically significant correlation (p<0.05) meaningfully contribute to the score.
Tier weight
Literature-backed weight (A=0.8, B=0.7, C=0.9, D=0.4) reflecting how directly each indicator type predicts TED burden, as established in the PMC review.
NOTE
The score is a relative ranking tool, not an absolute patient count. A score of 80 does not mean 80% of some benchmark — it means this prefecture ranks highly relative to all other prefectures on the combined indicator set. Use it to prioritise, not to size the market.
05
Quickstart — five steps
1
Orient with the KPI strip
At the top of the map panel, the five KPI cards show counts for all prefectures, High Priority, Medium, Divest, and Confirmed burden specifically. These are your headline numbers for the commercial team. Click any card to filter the map and ranking list to that segment instantly.
TIPClick the Confirmed burden card first — this is your most actionable cohort and the right starting point for rep deployment planning.
2
Read the map geographically
Scan the choropleth for geographic clustering. If several High Priority prefectures cluster in a region, that may allow one rep to cover multiple opportunities efficiently. Isolated High Priority prefectures in otherwise Divest regions may require dedicated resource decisions.
TIPUse the legend toggles to hide Medium and hide Divest — this shows only the High Priority prefectures on the map for a cleaner presentation view.
3
Use the ranking list for territory planning
The right panel lists all 47 prefectures sorted by composite score with a score bar. Use the search box to find a specific prefecture quickly. The score bar gives an instant visual of how far ahead the top-ranked prefectures are from those in the middle — that gap matters for resource prioritisation.
TIPIf score bars for ranks 1–8 are significantly longer than 9–15, that is a natural break point — the top 8 are meaningfully differentiated from the next tier and should receive primary resource allocation.
4
Drill into the detail panel
Click any prefecture on the map or in the list. The detail panel on the right shows: segment classification with rationale, the full indicator breakdown by tier as a bar chart, and two action blocks — one for field rep/MSL deployment and one for HCP education events. The indicator bars are normalised to the highest value in that prefecture, so you can see the relative signal strength within that market.
TIPLook at the Tier C indicator bars specifically. A Latent prefecture with a very low IVMP bar but a high Graves disease bar is a strong candidate for HCP education events — the disease burden is there, but treatment is not happening.
5
Filter by segment to plan action type
Use the legend sub-segment toggles to isolate Confirmed burden for rep deployment planning, Latent for HCP education event planning, and Surgical-heavy for oculoplastic specialist targeting. These three filter views map directly to the three commercial action types your field force can execute.
TIPWhen presenting to leadership, start with the map filtered to Confirmed burden only. It is the cleanest, most actionable view and gives the commercial team an immediate answer to "where do we go first."
06
Interpreting the results
The dashboard is a pre-launch opportunity map, not a post-launch performance tracker. Since X Brand launched in September 2024 and NDB data runs to 2023, every signal is a proxy for latent demand — not evidence of X Brand adoption. Interpret accordingly.
High score = high latent demand
A high composite score means multiple independent clinical indicators are signalling TED burden in this prefecture — Graves diagnoses, ocular symptoms, IVMP usage. The indicators reinforce each other through correlation, so a high-scoring prefecture is unlikely to be a false positive.
Not a X Brand adoption rate. Demand is currently unmet.
High IVMP = confirmed active TED
High IV methylprednisolone rates are the strongest proxy signal. IVMP is the Japanese standard of care for active moderate-to-severe TED — so high IVMP means a physician has already made the moderate-to-severe TED diagnosis and started treatment. These are the patients X Brand would replace IVMP for.
IVMP also used for other conditions — validate with NDB drug code specificity.
High surgery = late-stage population
Orbital decompression and strabismus surgery are irreversible interventions for patients whose TED progressed to cause permanent structural damage. A prefecture with high surgical volume and low IVMP had patients arriving at surgery without adequate earlier medical management — X Brand's prevention story is strongest here.
Surgical volume may reflect centre-of-excellence bias — one major hospital can drive a prefecture's numbers.
Latent = education gap, not absence
A Latent prefecture has high Graves disease and TRAb testing volumes — the upstream population is there — but low IVMP and low surgery. This is unlikely to mean TED doesn't exist. It more often reflects a local diagnosis gap, a weak endo-ophthalmology referral pathway, or limited specialist access. HCP education is the primary lever.
Validate with local KOL before concluding a diagnosis gap — claims data cannot capture what was never diagnosed.
Divest ≠ zero patients
Divest classification means the indicator profile is thin relative to other prefectures — not that TED is completely absent. Thyroid disease exists across Japan. A Divest prefecture may still have a few TED patients; it simply does not warrant dedicated field resource in the current cycle given opportunity cost elsewhere.
Revisit Divest prefectures annually as NDB data updates post-launch.
Score gap matters more than absolute rank
Rank 1 vs rank 2 is rarely commercially meaningful. What matters is the score gap between High Priority and Medium prefectures. If ranks 1–12 cluster at scores above 65 and rank 13 drops to 48, the natural investment boundary is between 12 and 13 — not at an arbitrary top-10 cutoff.
Use the score bars in the ranking list to visually identify natural break points.
07
Usage scenarios
Scenario A
Annual field force allocation — where to deploy reps and MSLs
A sales director is planning the next fiscal year's territory assignments and needs to decide which prefectures get dedicated headcount vs shared coverage.
Steps in the dashboard
1Click the High Priority KPI card to filter to the top tier prefectures.
2In the legend, click Confirmed burden sub-segment to isolate the switch-ready cohort.
3Look at the ranking list — note the score gap. Assign dedicated reps where scores cluster above the natural break point.
4Click each Confirmed burden prefecture to read the Field Rep action block in the detail panel. Note target specialties.
5Switch to Surgical-heavy — add MSL coverage here. These need scientific dialogue, not sales calls.
Expected output
Decisions enabled
›Prefecture list with rep deployment Y/N and priority level
›Target specialty by prefecture (endo-heavy vs ophth-heavy)
›Distinction between IVMP-switching accounts vs surgical-adjacent accounts
›Natural break point for headcount boundary (e.g. top 12 get dedicated coverage)
Scenario B
HCP education event planning — where to run disease awareness programmes
A medical affairs manager needs to select 6–8 prefectures for endocrinologist and ophthalmologist education events in the next two quarters.
Steps in the dashboard
1Click the Latent legend toggle to isolate prefectures with high upstream burden but low treatment signal.
2In the ranking list, note Latent prefectures by score — pick top 4–5.
3For each, open the detail panel. Check the Tier D co-visit bar — a low endo+ophth co-visit score confirms the referral pathway is broken and education is the right intervention.
4Also select 2–3 Confirmed burden prefectures for switching-focused events (different curriculum — these HCPs already know TED, just need X Brand data).
5Check geographic clustering on the map — can two nearby prefectures share a single regional event?
Expected output
Decisions enabled
›Short list of 6–8 prefectures with event type (awareness vs switching)
›Target audience per event (endocrinologists for Latent, ophthalmologists for Surgical-heavy)
›Identification of prefecture pairs that can share a regional venue
›Evidence rationale for each selection (indicator profile from detail panel)
Scenario C
Challenging a rep's intuition about their territory
A regional rep believes their prefecture should be High Priority based on anecdotal KOL relationships but the dashboard classifies it as Medium. How to use the data to investigate.
Steps in the dashboard
1Search for the specific prefecture in the ranking list search box.
2Click to open the detail panel. Read the segment rationale carefully — it explains specifically which signals are present and which are absent.
3Look at Tier C bars (IVMP, orbital surgery). If they are low but Tier A is moderate, the NDB is not showing treatment activity — even if patients exist locally.
4This could mean: (a) patients are truly thin, (b) a specialist centre in a neighbouring prefecture is absorbing patients, or (c) local IVMP is being administered in non-hospital settings not captured by NDB.
5Use this as the structured agenda for a KOL conversation — not to override the data, but to contextualise it.
Expected output
Decisions enabled
›Structured hypothesis for why NDB signal is low (not just "the data is wrong")
›Specific questions to bring to a local endocrinologist
›Decision: accept Medium classification, or flag for manual override with field evidence
›Check neighbouring prefecture — if it is High Priority, patient flow from your prefecture to theirs may explain the signal gap
Scenario D
Preparing a board-ready summary slide
A brand director needs to present the opportunity landscape to senior leadership in a 10-minute slot. They need one clean data story, not a full methodology briefing.
Steps in the dashboard
1Click High Priority KPI card. Note the headline count — this is your opening number.
2Turn off Medium and Divest using the legend toggles. The map now shows only the actionable prefectures in green/blue — take a screenshot of this view.
3Read the KPI cards: X Confirmed burden + Y Latent + Z Surgical-heavy = total High Priority. This gives the three-part narrative: switch, educate, prevent.
4Click the single highest-scoring prefecture and screenshot the detail panel as a worked example to show leadership what "High Priority" means concretely.
5Use the year tabs to show 2021 vs 2023 — if the signal is stable or growing, this strengthens the confidence in the classification.
›Slide 2: Three segment types — switch / educate / prevent — with counts and rationale
›Slide 3: One example prefecture detail panel — shows the evidence depth behind the classification
›Appendix: Full 47-prefecture table (export from ranking list)
08
Important caveats
DATA
Mock data is currently loaded. The current dashboard runs on synthetic data that matches the structural logic of the analysis but does not represent real NDB values. Replace mkData() in the dashboard script with output from phase3_correlation.py before any commercial decisions are made.
NOTE
No X Brand Rx data exists in the model. X Brand launched in Japan September 2024. NDB data runs to 2023. This is a pre-launch demand mapping exercise — the model identifies where patients exist, not where X Brand is being prescribed. Post-launch NDB data (available ~2026–27) should be used to update this analysis with actual adoption rates.
What the model can tell you
Where TED-associated clinical activity is concentrated · Which prefectures have active treatment burden · Where the treatment gap (high diagnoses, low treatment) is largest · Which prefectures have late-stage surgical burden · Geographic clustering for efficient field coverage
What the model cannot tell you
Exact number of X Brand-eligible patients · Whether low signals reflect truly thin markets or NDB coverage gaps · Patient-level treatment journeys · Why a physician chose IVMP over X Brand post-launch · How quickly a Latent market can be activated
CONFIDENTIAL: This prototype is intended for the Client X Brand Launch Team. The correlations shown are based on Signal 0 (PMC Literature) overlays on MHLW NDB Open Data. Reproduction or external distribution without consent is prohibited.