Skip to content

Smart Scheduling — Deep Research (external evidence · landscape · prior-art)

Date: 2026-06-03 · For the 6/04 patent working meeting. Complements the internal field evidence (Smart_Scheduling_Field_Evidence). Headline: the published literature validates our clinical design — and the prior-art landscape narrows where our real novelty is. Better to know that before counsel than after.

1 · Clinical evidence — it backs our design

  • Multi-component / tailored beats single reminders. Systematic reviews of older-adult adherence find the effective combination is education + regimen simplification + remote follow-up + caregiver support + tailored (AI) reminders — especially in polypharmacy/multimorbidity. A lone reminder is the weak version. → validates our learned + cross-engine + Circle approach over a dumb alarm.
  • Reminders alone help — but only for a limited time, and not for every cause. Reviews: text/phone reminders boosted adherence in ~30% of studies but are "only helpful for a limited time," and are "likely ineffective if the issue is belief/trust, not memory." → direct external confirmation of our "fatigue, not forgetting" finding and our adaptive de-escalation (remove nags as a routine proves out).
  • Two-way (confirmation) is the recognized step up. The literature explicitly flags uncertainty between 1-way (remind only) and 2-way (remind + confirm intake) — the open question is exactly the lane we're entering. → validates confirmation-first.
  • "Drug forgiveness" is established science — and it's our tolerance window by another name. Longer half-life = more forgiving of a missed dose; short half-life demands strict timing. Crucially, forgiveness is a function of pharmacodynamics too, not just half-life (aspirin: 2–3 h half-life but 7–10-day action). → our drug-class tolerance window is grounded in a named clinical concept; the PK + PD nuance is a sophistication point (and a reason it must be clinically curated, not naive).
  • Missed-dose handling is an acknowledged unsolved problem. A systematic review of patient leaflets/SPCs found missed-dose guidance is inconsistent/absent. → real white space for "what do I do at the dose, and if I miss it."

2 · Competitive landscape

  • Real market: medication-reminder apps ≈ $0.5–1.2 B (2024–26), ~8.5–13% CAGR.
  • Players: Medisafe (drug-interaction checker; moved to paywall Jan-2026), Hero (smart dispenser hardware — 90-day/10-med, missed-dose + refill + caregiver alerts, Travel Mode, adherence reports, ~$30–60/mo), Pillo / MyTherapy / Roundhealth (free reminders + symptom/vitals + streaks/behavioral psychology).
  • Trend: personalization + AI-driven tailored reminders; smart-pill-bottle connectivity expected in ~35% of apps by 2027.

3 · ⚠ Prior-art reality check (for FTO — read before counsel)

Several things we framed as "novel" are heavily prior-arted: - Pre-emptive miss prediction → ML adherence prediction is a mature field: AUC 0.75–0.85 for 7–30-day non-adherence, multiple papers (Nature Sci Reports; JMIR scoping review), and reinforcement-learning reminder schedulers already published. Not novel on its own. - Learned / adaptive / context-aware reminder timing → exists; "context-aware reminders" already report 92%+ adherence. Not novel alone. - Caregiver notifications · two-way confirmation · streaks → all shipping today (Hero, Medisafe, Roundhealth).

Implication: we should not lead the patent on "we predict misses" or "we personalize reminders" — that ground is crowded.

4 · So where IS our defensible space (sharpened)

Claim the combination + the genuinely distinctive pieces, not prediction/personalization in isolation: 1. Drug-forgiveness-based tolerance built into the reminder behavior — a weekly drug nags differently than a daily one, curated by PK + PD (not just half-life). Few consumer apps do clinically-curated tolerance. 2. Dose-time context carried with the schedule (food/timing rules) — the unsolved white space from §1. 3. Cross-engine "medication-change" event with the fall barometer — the integration is ours alone; no competitor pairs scheduling with a fall-risk engine. 4. Provenance-aware (pharmacist-verified) scheduling — most apps are self-entry; the pharmacy-verified gold-provenance path is distinctive. 5. Wellness-lane framing as the design — reminders/confirmations, never dosing instruction.

→ The patent's strength is the system integration (scheduling × fall engine × pharmacist provenance × clinically-curated forgiveness), not any single learning trick.

5 · What this changes for tomorrow

  • Reframe the deck's "novel elements" to lead with the distinctive combination (above), and present prediction/personalization as table-stakes we do well, not the claim. (Recommend — your call.)
  • FTO ask to counsel: the ML-adherence-prediction + RL-reminder space is crowded — search it specifically. Our claim should sit on the integration + clinically-curated forgiveness + provenance.
  • Clinical citations strengthen S2/S3 (evidence of need) and give the tolerance concept a literature anchor ("drug forgiveness").

Sources

Clinical evidence - Multilevel Interventions to Improve Medication Adherence in Older Adults (systematic review/meta-analysis, 2026) - Enhancing Medication Adherence in Older Adults — evidence-based strategies (JAGS 2026) - Effects of medication adherence interventions for older adults with chronic illnesses (meta-analysis) - Text-message reminders & adherence in T2DM (meta-analysis) - Effectiveness of mobile apps on medication adherence (systematic review/meta-analysis) - Drug forgiveness and patient adherence (AARDEX) - A pharmacokinetic and pharmacodynamic analysis of drug forgiveness (PubMed) - What should patients do if they miss a dose? Systematic review of PILs/SPCs

Landscape - Medication reminder apps market report - Medisafe · Hero smart dispenser

Prior art / AI scheduling - Machine Learning and Medication Adherence: Scoping Review (JMIR) - Predicting medication adherence with ensemble/deep learning (Nature Scientific Reports) - AI-Powered Medication Reminder using Reinforcement Learning