Pattervia

DECISION INTELLIGENCE FOR DIGITAL HEALTH

Decide what to deliver to each user—and when.

Predefined rules map user scores and risk levels to actions: low engagement → send a reminder; high risk → flag for review. But adapting those actions to each person can still leave your team updating rules, reviewing alerts, and tailoring support manually.

We’re building Pattervia to learn from behavioural patterns and past intervention responses—helping your team decide what support to offer each user and when, with less manual decision work.

Open to design partners See an example
FROM DATA TO DECISIONS
Longitudinal data flows through measurement, decisions, delivery, outcomes and evaluation. Research measurements also inform evaluation, and evidence feeds back into decisions. A text description follows.
Read the text description
  1. Longitudinal data: Wearables, symptoms, clinical records and interaction history feed into measurement and interpretation.
  2. Measurement and interpretation: Features, personal baselines and state estimates inform decisions.
  3. Decision: Who needs what support, and when? This informs delivery.
  4. Delivery: A coach, clinician, app or treatment device delivers the intervention.
  5. Observed outcomes: Behavior, symptoms, function and resource use feed into evaluation.
  6. Evaluation: Did the intervention cause an improvement?

A separate branch connects measurement and interpretation to research measurements—biomarkers and trial endpoints—which also feed into evaluation. A dotted feedback arrow returns from evaluation to decision, labeled “Evidence to improve decisions.”

From data to decisions.

Scroll to explore the full diagram. Open the image separately

The complete workflow, including the research measurements branch and evidence feeding back into decisions.

Pattervia · Decision pipeline

From signals to the next step.

Maya · demo
Click a stage to explore

Step 3 of 5: Choose an eligible next step

STEP 03 · PATTERVIA

Choose an eligible next step

Combine context, response estimates and programme constraints into a recommendation for review.

Suggested action

Evening support message
18:00 local · approval pending

  • Check-ins remain outstanding
  • Timing matches Maya’s preference
  • No coach capacity available

48-hour completion · illustrative estimates

No extra outreach
20%
Standard reminder
25%
Evening support
45%
Recommendation outputJSON
{
  "decision_id": "001",
  "person_id": "maya",
  "eligible_actions": [
    "no_extra_outreach",
    "standard_reminder",
    "evening_support_message"
  ],
  "excluded_action": {
    "action": "offer_coach_call",
    "reason": "no_coach_capacity"
  },
  "outcome": "checkin_completed_within_48h",
  "illustrative_probabilities": {
    "no_extra_outreach": 0.2,
    "standard_reminder": 0.25,
    "evening_support_message": 0.45
  },
  "recommendation": {
    "action": "evening_support_message",
    "scheduled_for": "2026-09-08T18:00:00+12:00",
    "approval": "pending",
    "delivery": "not_sent"
  },
  "estimate_basis": "illustrative_assumptions"
}

Research behind the approach.

Pattervia is being developed by Adi Choi, a final-year PhD student researching how longitudinal behavioural data can reveal individual patterns and changes over time.

Her publications include:

  1. npj Digital Medicine

    Personalised modelling of routine variability and affective states

    Adrien Choi, Danielle Lottridge and Jim Warren

    Read the paper

  2. JMIR mHealth and uHealth

    Digital Phenotyping for Stress, Anxiety, and Mild Depression: Systematic Literature Review

    Adrien Choi, Aysel Ooi and Danielle Lottridge

    Read the review

These publications inform the approach. Our first evaluations will test whether that research translates into useful support decisions. Our longer-term aim is to help digital care teams use longitudinal context to make support more responsive across conditions.

Looking for design partners

We’re looking for digital care providers with an existing coaching or support team who want to improve program completion.

Together, we’ll evaluate one decision: which participants need a check-in, and when?

What you bring

  • A defined participant cohort
  • Historical participation and support records
  • Someone who understands the workflow

What you receive

  • A comparison against your current approach
  • Example participant reviews
  • A recommendation on whether a live pilot is justified

The partnership starts with a scoped evaluation. We’ll test whether changes in individual patterns add useful information beyond your current review process.

Discuss an evaluation

Email Adi to discuss your cohort, workflow and evaluation scope.

adi.choi@pattervia.nz