Plaiatech

MVP Delivered

Replacing paper intake and hallway assessments with AI-powered pre-surgical screening

Next.js
Node.js
TypeScript
AWS Lambda
DynamoDB
SQS
SNS

The Setup

Before any surgery, an anesthesiologist needs to evaluate the patient — airway anatomy, cardiac history, current medications, allergies, prior reactions to anesthesia. It's a detailed clinical assessment that determines how safe it is to proceed.

The problem is timing. In most clinics, this evaluation happens the day of surgery or requires a separate in-person visit days before. If something unexpected shows up — a drug interaction, an uncontrolled condition — the surgery gets postponed. Same-day cancellations waste operating room time, stress patients, and cost the clinic money.

A professional with experience in the field approached me to fix this: move the entire pre-anesthesia evaluation to the patient's phone. No app download, no account creation — just a link that gets them through a clinically valid assessment from home.

The Constraints

Three constraints shaped every technical decision:

Clinical accuracy is non-negotiable

The system scores patients on established medical scales used in clinical practice. These aren't suggestions; they're the same standards anesthesiologists use in person. Getting them wrong means patient safety is compromised.

Patients aren't tech-savvy

The typical user is a 65-year-old facing surgery, not a digital native. They get an SMS link, tap it, and need to complete everything in their phone browser. No app store, no login, no second chances to figure out confusing UI.

Every clinic does it differently

An orthopedic clinic asks different questions than a cardiac surgery center. Each specialty has its own protocols, and clinics need to customize intake flows without calling a developer every time.

What I Built

Each piece of the system maps back to a constraint. Here's how the solution came together.

Solving for: patients aren't tech-savvy

Patient Experience

The patient receives an SMS with a link. No app to download, no account to create. They tap the link and land on an adaptive questionnaire organized into blocks — each screen presents multiple related questions, with dependent questions that show or hide based on previous answers. The entire experience is optimized for a phone screen — large touch targets, clear language, and a logical flow that guides patients through sections like medical history, medications, and document uploads.

9:41
Interactive DemoCardiovascular · Step 1/5

Cardiovascular

Does the patient have any cardiovascular history?

Solving for: clinical accuracy is non-negotiable

AI-Powered Clinical Assessment

As the patient works through the questionnaire, each answer is tagged with a specific clinical score. The system collects responses across many dimensions — medical history, current conditions, previous surgeries, and more — and an algorithm evaluates all answers together to produce a clinical risk assessment. One input among many is a photo analyzed by a computer vision model to determine a clinical classification relevant to the procedure.

The algorithm combines all these scored inputs to compute an overall risk classification. The anesthesiologist receives a pre-filled risk assessment instead of starting from scratch — with every contributing factor traceable back to a specific patient answer.

Solving for: every clinic does it differently

Questionnaire Builder

Each clinic configures its own intake flow through a visual builder. Questions can be reordered, branching logic connects blocks so different patient profiles trigger different follow-up paths — and none of it requires a developer. A cardiac clinic can add deep questions about pacemakers while an orthopedic clinic focuses on joint replacement history.

app.example.com/admin/questionnaire-builder

Medical Staff Dashboard

Everything consolidates into a single dashboard for the anesthesiologist. Patient responses, AI-generated risk scores, parsed medications, uploaded documents — all reviewed before the patient ever arrives. The system auto-generates a structured pre-anesthesia report that would have taken 30 minutes to compile manually.

app.example.com/dashboard
DashboardPre-Anesthesia Evaluations
4
Total
2
Completed
1
In Progress
PatientDateStatus
María García2024-11-18
Completed
Carlos López2024-11-19
In Progress
Ana Rodríguez2024-11-20
Pending
Jorge Martínez2024-11-20
Completed

Architecture

Fully serverless. Patient requests hit API Gateway, which routes to Lambda functions handling questionnaire logic, AI assessment, and document processing. DynamoDB stores patient data with strict tenant isolation — shared DynamoDB tables use clinic-scoped partition keys, with access control enforced at the API layer. S3 handles document storage with pre-signed URLs for secure upload and retrieval.

Frontend
Next.js
TypeScript
Tailwind CSS
API
Node.js
AWS Lambda
API Gateway
Data
DynamoDB
S3
Messaging
SQS
SNS
Ops
CloudWatch
Serverless Framework
GitHub Actions

Results & Impact

~15 min
Average evaluation time
0
In-person visits required
Metric TBD
Metric TBD

The MVP has been delivered and is entering its first round of real-world testing with clinics. The next iteration will be shaped by how medical staff and patients respond to the system in practice.

Tech Stack

Next.js
Node.js
TypeScript
AWS Lambda
DynamoDB
S3
API Gateway
SQS
SNS
CloudWatch
Serverless Framework

Let's talk

Have a project in mind or just want to chat? Reach out.

Get in touch