Braincheck24
Public demo · Library · Health · Reflux

Reflux100: What helps my heartburn?

Property of Prof. Paolo Brenner

What helps my heartburn? Set symptoms, warning signs, triggers, treatment and goals. The newly designed Reflux100 network shows which urgency, likely patterns, checks, topics for your doctor and everyday steps it links to them. Warning signs such as chest pain, vomiting blood and swallowing problems always lead to an emergency call, the emergency department or a doctor.

Important. Free, restricted demo. Learning simulation with pattern data, not a diagnosis, not a treatment recommendation, no substitute for a medical examination. Values are normalised activations from 0 to 100 %, not probabilities. All results are suggestions only, without guarantee.

No diagnosis, no therapy. Synthetic learning network built from self-defined pattern rules, not trained on patient data and not clinically validated. In chest pain spreading out, breathlessness or sweating call 112 at once – it may be a heart attack. In vomiting blood or black stool go to the emergency department at once. Have swallowing problems or unintended weight loss checked soon.

The network (100 → 60 → 100 neurons, ReLU / Sigmoid) was newly designed by us and trained on 1,000,000 synthetic rule-based cases (held-out data: R² 0.999, RMSE 0.003). There is no real case or market data and no expert validation. The demo uses 40 inputs, 40 outputs (excerpt from 5 categories) and 8 example cases; the other 60 inputs are fixed at 0 %. The bars start at 0.

0 of 40 inputs active

Your details

Drag bars (0–100 %) · click a name to switch fully on / off

Linked model topics

Activation 0–100 %

Strongest links (by weighting)

    Benefits of the full version

    FeatureDemoFull version
    Inputs (input neurons)40all 40 in 5 categories
    Outputs (output neurons)40all 40 in 5 categories
    Example cases8 as excerptall 8
    Ranking by weighting40 outputsall 40
    Scenario link and printyesyes, for all 40 inputs
    EditionsGerman and EnglishGerman and English, usable offline
    UseViewLicence with personal key

    The full version is also a learning simulation without guarantee.

    Full version at braincheck24.info

    How it works

    Each input is a draggable bar from 0 to 100 %. Values pass through 60 hidden neurons (ReLU) to the outputs (sigmoid). “Sort outputs” ranks all outputs by weighting, strongest first.

    What the numbers mean

    An activation is a relative value from the embedded weights on a 0 to 100 % scale, not a probability and not a recommendation for an individual case.

    What helps day to day

    Smaller evening meals, no food three hours before bed, raising the head of the bed, less alcohol, coffee and smoking and losing weight help many people. Warning signs always take priority in the network.

    Patent application & rights

    The neural visualisation unit used here is the subject of European patent application EP 26000055.9 (filed 17 July 2026, patent pending). Interface, weights, labels and texts are protected by copyright; copying or commercial use only with written consent. Licences: braincheck24.info.

    © 2026 Prof. Dr. med. Paolo Brenner · Braincheck24. All rights reserved. No guarantee for content or results; outputs are suggestions only.

    Scenario link