Where a verified dispense is worth the most

01.1

Drug Group Framework: Segment, Qualify, Prioritise

A Repeatable Screen, Not a Fixed List: 24 Drugs to 28

Three stages — what a missed dose costs, four mechanical gates, then whether it is worth tooling.

1. 1

   Stage 1 · Segment

   By what a missed dose costs, and who therefore pays: A therapy loss (pharma first), B acute event (pharma on patent, plan once generic), C aggregate cost (plan-funded).

2. 2

   Stage 2 · Qualify — four gates, all must pass

   Form: a tablet, or a non-gelatin capsule. Container: the patient receives the maker's bottle, not a blister wallet. Geometry: one opening size and one shape as the patient titrates and across suppliers. Regimen: chronic or cyclic with a defined plan — not PRN, not split tablets.

3. 3

   Stage 3 · Prioritise

   A qualified drug is not automatically worth tooling: avoided cost per patient-year, patients on the molecule, who funds and for how long, moulds × sizes, and who signs.

U.S. patient counts: ClinCalc DrugStats (MEPS 2023) where listed, else Coledy sizing. Packaging: FDA NDC Directory. Shape and size: DailyMed. Running the gates is what removed four widely-prescribed specialty drugs from an earlier version of the tables that follow.

01.2

A Dose Is Not Always One Tablet

All 28 regimens read off FDA labelling — and the three things it changed.

What the audit did not change twenty-one once-daily and seven twice-daily, confirmed against all 28 labels — the split already published, so no figure moved. What it did change the geometry gate now routes a split dose to a device per bottle instead of rejecting it, and pPDC caps per bottle.

What has not been measured: a four-tablet dose is four separate tilt-and-release actions. Field use so far has been on one-tablet regimens, so whether patience holds across four tilts is untested on five of the highest-value rows.

Regimen is tablets per dose × doses per day, read off FDA labelling for all 28 rows; shares round to 101%. Loading and ramp-up doses are named separately — Brilinta and Plavix load at two and four tablets, Eliquis opens DVT/PE at two, Venclexta ramps on a wallet.

01.3

The Screen Rejects More Than It Accepts

Every gate is run against FDA labelling — here is what failed, and the line that failed it. A list of drugs is a claim; a list of what failed and why is a method.

Four of five permanent exclusions fail on form or regimen, not geometry. Geometry decides among generics — and there it is usually recoverable, because what fails is that suppliers disagree on shape. The same lines that reject a molecule are the spec a partner can design to.

02.1

28 Priority Therapies, Three Groups

Six therapeutic areas, three entry groups — nine, eleven and eight drugs, screened from U.S. prescription data.

Two funding routes, never zero: a manufacturer whose therapy is lost when a dose is missed, or a payer already scored on the measure. A drug with neither is not on this list — that is what the screen tests for, and why the list is 28 and not 2,800.

Counted per drug — a patient treated for three of these appears three times, and carries three devices. Source: ClinCalc, CDC, FDA.

02.2

Group A — Therapy Loss

High-cost therapies where a missed dose lets resistance or progression in. ~0.83M U.S. patients.

Patients: ClinCalc, CDC, FDA. U.S. basis — generics at CMS NADAC plus dispensing fees, branded at list price. Avoided cost per failure event, from the sources on the next table.

02.3

Group B — Acute Event

A missed dose ends in relapse, admission or a costlier therapy. Mostly generic — paid for by the plan. ~3.2M U.S. patients.

Patients: ClinCalc, CDC, FDA. U.S. basis — generics at CMS NADAC plus dispensing fees, branded at list price. Avoided cost per failure event, from the sources on the next table.

02.4

Group C — Small per Patient, Enormous in Total

Cheap, widely used tablets where adherence drives Medicare Star Ratings. ~51M U.S. patients. ★ = Star Ratings measure drug.

★ marks drugs that count toward Medicare Star Ratings adherence measures — the reason a payer pays for a low-cost generic.

Patients: ClinCalc, CDC, FDA. U.S. basis — generics at CMS NADAC plus dispensing fees, branded at list price. Avoided cost per failure event, from the sources on the next table.

02.5

Where Each Avoided-Cost Figure Comes From

Avoided cost is what one failure event costs when a dose is missed. Each figure is a published health-economics result — not a Coledy estimate.

Drug cost on an acquisition basis: CMS NADAC per unit × 365 plus twelve dispensing fees for generics; list price for branded rows. Aubagio uses median generic WAC.

03.1

Real-Time Evidence to Stop Controlled Drug Diversion

Legacy tracking stops at the pharmacy counter. Coledy verifies the dose after it — at the dose level.

• EMR · Rx issued
• Claims · pickup
• Auth · patient phone
• Verify · GPS location
• Evidence · single-tablet log
• Outcome · anomaly alert

Coledy layer: GPS verification + single-tablet log turn the untraceable dose event into settlement-grade evidence.

Regulatory A top-priority issue for the DEA; FDA imposes REMS obligations on pharma — a clear, mandated payer.

Traction: early interest from government / regulators at U.S. expos (pre-pilot, not yet executed). Source: CDC WONDER & SAMHSA (2024–2025 basis).

03.2

Obesity Therapy: Capturing the Oral GLP-1 Market

Two approved pills in fifteen months and a phase 3 field behind them. Every one of them is a tablet in a bottle.

The moisture cap is a part, not a barrier. Oral semaglutide keeps its drying agent in the cap. On a round bottle the Coledy build is an add-on closed by a single child-resistant cap of Coledy's design — so the desiccant and the seal are part of that design. Not yet demonstrated: equivalent moisture performance.

Approvals: FDA. Phase status as most recently reported by each company. Not in the financial model — this widens an existing route, it does not add a forecast line.

04.1

Healthcare AI: Ground-Truth Adherence Data

A predictor measured with error drags its own coefficient toward zero — and adherence is measured with error everywhere.

A verified dispense raises λ on the adherence variable toward 1. That does not make a model smarter — it removes the attenuation that was hiding the effect. This holds for non-differential error; the systematic over-reporting in self-report shifts the mean and needs separate treatment.

Spearman, Am J Psychol 1904 (correction for attenuation); Carroll, Ruppert, Stefanski & Crainiceanu, Measurement Error in Nonlinear Models, 2nd ed., 2006; Frénay & Verleysen, IEEE Trans Neural Netw Learn Syst 25(5), 2014.

04.2

Healthcare AI Trains on Inferred Adherence, Not Verified

A fill, an opening and a memory are all inferences — a verified dose never reaches the model.

• It cannot be bought or backfilled

04.3

Predictive Value from Adherence Data

Precision dispense data fused with partners' lifelog signals is designed to train models that predict risk and generate real-world evidence.

1. 1

   Data fusion

   Fuse verified pPDC with partners' lifelog data — fill the clinical-data gap with high-resolution behaviour.

2. 2

   Model development

   Learn individual patterns to flag adherence drop-off risk; use adherence–biomarker correlations to predict complications.

3. 3

   AI on adherence

   Analysis on measured adherence strengthens clinical development and regulatory readiness.

Not in the forecast — no data-licensing contract exists.

04.4

Digital Health Networks: Daily Social Network Engagement

Platforms Bring Reach; Coledy Brings a Verified Dispense Record

A scheduled, device-triggered event every day — engagement the platform does not have to manufacture.

• A daily reason to open the app
• A care circle worth subscribing for
• Health content it can stand behind

Consent first Anything shared beyond the patient — a carer, a circle, a feed — is opt-in per recipient and revocable. Adherence is health data, not a status post.

04.5

Supplements Claim Efficacy With No Proof of Dosing

Four categories sold on a dosing regimen — and none can show the regimen was followed.

Regimens are the manufacturers' own label directions. A verified record that the regimen was dispensed is the missing half.

05.1

The Channel That Needs No Market Entry

An interventional trial supplies its own bottles — so the dispensing format of the country the patient lives in does not apply.

R&D waste figures: Tufts CSDD and Watanabe (2018). A post-authorisation study on a marketed drug is different — the patient collects an ordinary pack, so that half needs the valve-less Universal build and a bottle market. No clinical contract has been signed.

05.2

Connected-Care Infrastructure for B2B & B2G

The dose-data layer scales — from the individual to community care to smart cities.

Data flow: individual › community › urban infrastructure › policy. Multi-sector B2B & B2G value creation across pharma R&D, clinical data, and global platform expansion.

Illustrative — no counterparty approached.

05.3

Seven B2B Channels, and How Each One Signs

The go-to-market plan targets seven B2B channels across pharmacies, hospitals, distributors and digital platforms.

• Remote and online pharmacy
• Local pharmacy network
• Hospital partnerships
• Global partnerships
• Pharma distribution
• Caregiver network
• Digital channels

These are target channels for the go-to-market plan. None is contracted today — Coledy intends to sell into them; it does not already operate them.