Research Repository

A Signal That Precedes the Swelling

Lymphoma Leaves Inflammatory and Cellular Markers in the CBC Before Diagnosis—the Algorithm Remains to Be Built

Already in the Data

Kidney Cancer Leaves a Detectable Blood Signature Months Before Diagnosis, Yet the Algorithm Has Not Been Built

Beyond PSA

Prostate Cancer Leaves a Distinct Pattern in Routine CBC and Chemistry—Machine Learning Improves on PSA Alone

Cancer Detection Summary Table 20260322 V2

Copper Damage Leaves a Distinctive Blood Fingerprint

Wilson Disease Produces a Recognizable Liver Enzyme Pattern That Machine Learning Can Flag Years Before Neurological Injury

Executive Summary

FDA Regulatory Exemption for CBC-Based AI Screening Algorithms

From Blood Draw to Clinical Action

How Maccabi and Geisinger Deploy Early Detection Algorithms, What the Blood Tests Cost, and How to Scale This Nationally and Globally.

Hidden in Plain Sight

Routine Clinical Blood Data Contains a Lung Cancer Signal, and a Machine Learning Algorithm Has Already Learned to Read It

Introduction

Iron Overload, Written in Your Blood

Haemochromatosis Leaves a Decades-Long Metabolic Signature That Machine Learning Reads with 94% Accuracy

Noncancer Disease Summary Table Landscape 20260322

Overview

Siemens Healthineers as REDI's Primary Strategic Partner: The Case and the Playbook

Sixty Days Before the Collapse, the Blood Already Knows

Declining Hemoglobin, Rising RDW, and Falling Sodium Precede Heart Failure Hospitalization—Machine Learning Reads the Trajectory with AUC Up to 0.93

The Blood Already Knows

How a Machine Learning Algorithm Built on Routine Blood Tests Has Already Proven It Can Catch Colorectal Cancer Months Before Diagnosis

The Blood Already Predicts the Heart Attack

Routine CBC and Metabolic Panels Contain a Years-Long Cardiovascular Signal That Machine Learning Reads Better Than Framingham

The Blood Behind the Bladder

Bladder Cancer Leaves Pre-Diagnostic Signals in Routine Blood Tests Months Before Diagnosis—Machine Learning Achieves AUC 0.92

The Blood Has Been Warning You for Years

Five to 15 Years Before Diagnosis, Routine Metabolic Panels Already Show the Trajectory Toward Type 2 Diabetes—Machine Learning Reads It with AUC 0.90

The Blood Panel Predicts Thyroid Failure Before the TSH Does

An XGBoost Algorithm Detects Hypothyroidism with AUC 0.91 Using Only Routine Lab Tests—No Thyroid Function Tests Required

The Cancer the CBC Was Made to Find

Leukemia's Pre-Diagnostic Signal in Routine Blood Counts Is Measurable Years Before Diagnosis, and Machine Learning Reads It with 92 to 98% Accuracy

The CBC Already Signals the Infection Before the Doctor Does

Three Machine Learning Algorithms Are Deployed in United States Hospitals—TREWS Cut Sepsis Mortality by 18.7% in a Prospective Multi-Site Study

The Cholestatic Fingerprint No Algorithm Has Learned to Read

PBC Leaves a Distinctive Alkaline Phosphatase Signature in Routine Metabolic Panels, and No Machine Learning Algorithm Has Been Built to Find It

The Electrolyte Signal That Nobody Has Built an Algorithm to Read

84% of Addison's Patients Have Hyponatremia at Diagnosis—a Canine Machine Learning Algorithm Already Achieves 99% Accuracy on the Same Biochemistry

The Fibrosis Is Already in the Blood Work

Machine Learning on Routine Metabolic and Hematological Panels Detects Advanced Liver Fibrosis with AUC 0.91—Better Than the Clinical Tool Already in Use

The First Deployed Machine Learning Algorithm From Routine Labs

Klinrisk—CE-Marked in 2025 and Validated on 4.8 Million Patients—Predicts Kidney Failure Years Before Creatinine Triggers an Alert

The Inflammation Signature

Breast Cancer Produces a Measurable CBC Inflammatory Pattern Before Diagnosis—Machine Learning on 396,000 Women Has Already Proven It

The Last Mile: Why Patient Adherence Determines Whether Early Detection Saves Lives

The Lipid Panel Already Knows

Familial Hypercholesterolemia Leaves a Distinct Routine Blood Signature That Machine Learning Reads with 90%+ Detection Accuracy, and Deployed Algorithms Already Prove It

The Liver Announces Itself

Routine Blood Chemistry Contains an Early Hepatocellular Carcinoma Signal, and Machine Learning Has Already Learned to Read It

The Most Readable Signal of All

Multiple Myeloma Announces Itself in the CBC and CMP Years Before Diagnosis—the Blood Signature Is Among the Clearest in Oncology

The Quiet Malignancy with a Loud Blood Signal

Ovarian Cancer's Pre-Diagnostic Platelet and Biomarker Pattern Is Detectable Long Before Symptoms—Machine Learning Reads It with 96% Accuracy

The Silent Killer That Speaks Early

Pancreatic Cancer Announces Itself in Routine Blood Chemistry Two to Three Years Before Diagnosis—the Algorithm Exists in Research, Not in Practice

The Stomach Speaks Through the Blood

Gastric Cancer Produces a Detectable CBC and CMP Signature Before Diagnosis—the XHGC20 XGBoost Algorithm Achieves AUC 0.91

Using INSIGHT at Weill Cornell Medicine to Build Early Detection Algorithms

A practical assessment of data availability, scientific workforce, costs, and timeline for developing machine learning algorithms across 25 cancers and diseases.