Evaluations that produce data, not just paragraphs.
When you perform a guided evaluation in MyTherapyWizard, you're not filling in a narrative template. You're capturing scored performance in structured domains.
The evaluation itself knows which items measure what, visual perception, motor planning, receptive language, safe mobility, executive function, whatever the instrument is designed to assess. When the evaluation is complete, you don't have a narrative report that needs to be interpreted. You have scored data that the rest of the system can use.
You know which domains the client performs within expected range, which domains fall below competence, and where the baseline sits within each affected domain. That profile isn't just something you read, it's something the rest of the system reasons from.
Plans of care that reason from the evaluation.
Here's where most EMRs hand everything back to you. You just evaluated the client; now write a plan of care. What goals? In what order? At what level?
MyTherapyWizard generates the plan of care directly from the evaluation data. It identifies which domains the client scored below competence on. It reads the baseline performance within each of those domains, not just "deficit yes/no" but where the client actually is. And it sequences goal suggestions by clinical hierarchy, the order in which skills typically develop or rehabilitate, which foundational skills scaffold more complex ones.
In plain terms: the system won't suggest targeting handwriting legibility if the client doesn't yet have the underlying visual-motor foundations. It won't suggest complex sequential tasks if single-step following is inconsistent. It won't suggest high-level community mobility goals before safe transfers are established. It suggests the goals that are clinically appropriate right now, for this client, at this stage of their care.
That's the kind of reasoning that usually takes ten years of clinical experience to develop reliably. New therapists often write goal sets that look fine on paper but don't build on each other, and patients make slower progress as a result.
MyTherapyWizard encodes the scaffolding logic directly, so every POC coming out of the system is clinically coherent, whether the therapist is a new grad or a twenty-year veteran. You remain the clinician. Every suggested goal is reviewed and edited by you. You add, remove, modify, override. The system is doing the reasoning the way a strong supervisor would, giving you a thoughtful starting point based on the data, not replacing your judgment. But the starting point is informed, not blank.
Documentation that knows who's reading it.
Before the POC is generated, you select the audience: medical or educational. That choice changes everything downstream.
Language payers require. Centers medical necessity. Frames goals around deficits, skilled intervention, and functional impact. For insurance reviewers, Medicaid auditors, physicians signing orders.
Language educators require. Centers educational necessity. Frames goals around strengths, access, and participation. For IEP teams, school-based providers, educational agencies.
The POC comes out pre-tuned to the reader it's meant for, with a medical necessity statement for medical audiences, or an educational necessity statement for educational audiences, built in from the start. No more retrofitting language to match your audience after the fact.
Goals that are both readable and measurable.
When you write or edit a goal in MyTherapyWizard, you're composing structured components, a verb, an object, a qualifier, a measurement type (accuracy, frequency, rate, duration, completion), a target value, a support level, a mastery rule, a baseline. The system assembles these into a natural-language goal that reads exactly like a therapist wrote it.
But underneath the readable sentence, the goal is machine-queryable. The system knows that the target is 70% accuracy, the support level is independent, the mastery rule is four out of five trials, the baseline was 25%. That structure is what makes everything else possible.
Goals are suggested by our AI assistant, but the AI isn't guessing. It's working from the merged evaluation data associated with the POC, real scores, real baselines, real domain profiles, and sequencing suggestions by developmental priority. The AI is faster than a therapist writing from scratch, but it's also more accurate, because it's not inventing the clinical picture. The clinical picture is already in the evaluation data.
Sessions that capture measurement, not prose.
Fifteen seconds per goal. Correct out of attempted. Cues level. Level of assistance. Any goal-specific challenges that showed up (attention, regulation, fatigue).
From those inputs, the system calculates a Goal Performance Index (GPI) for each goal, a number that combines how well the client performed against the target with how much support was required to perform at that level. It's not a therapist's gestalt rating. It's a calculation from the actual data the session produced.
Why a gestalt rating isn't enough
A client scoring 70% accuracy with maximum assistance produces a lower GPI than a client scoring 70% accuracy independently, because those are clinically different events, and a measurement system that can't tell them apart isn't measuring what therapy actually does. Ours can. Ours does.
The Session Performance Index (SPI) averages GPI across the goals addressed in a single session. Across sessions, GPI values form a trajectory for each goal.
Progress reports that compose themselves.
Over time, GPI and SPI produce trend data: this goal went from 15% to 45%, that one from 15% to 30%, the overall SPI improved by 23 points. When it's time to write a progress report, the system has already done the measurement work.
The system composes a narrative report that reports the measured changes, interprets them in clinical context, identifies areas of continued need, and generates an audit-defensible medical necessity statement, all grounded in the structured data underneath. If data is limited, the report says so. If progress is ambiguous, the report reflects that. If progress is meaningful, the report documents it with numbers a payer can verify.
This is the difference between AI that generates plausible-sounding clinical prose and a reporting system that generates accurate clinical prose, because it's working from data that's already been measured.