Machine learning methods to predict antibiotic resistance ... · Machine learning software to...

Machine learning software to predict antibiotic resistance and

pathogen typesDr. Sebastian Dümcke

Clemedi AG, Schlieren, Switzerland

Advancing Data Technologies to corner AMR June 5th, Amsterdam

Image credits: Microrao; https://commons.wikimedia.org/wiki/File:Antibiotic_resistant_bacteria.jpgImage credits: Netha Hussain; https://en.wikipedia.org/wiki/File:Infiltrating_ductal_carcinoma_20X.jpg

Image credits: Microrao; https://commons.wikimedia.org/wiki/File:Antibiotic_resistant_bacteria.jpgImage credits: Netha Hussain; https://en.wikipedia.org/wiki/File:Infiltrating_ductal_carcinoma_20X.jpg

Diagnostic Platform

0 h 6-24 h

IVD Kit IVD Software

(1)

DNA/RNA

(2) (3)

Amplification Sequencing

(4)

Machine-learning

(5)

Diagnostic report

OPTIMAL INITIAL ANTIMICROBIAL THERAPY

This translates to

better patient care

reduced hospitalization costs

control the spread of antimicrobial resistance

A new standard for in vitro diagnostics

FAST COSTAUTOMATEDCOMPREHENSIVE

100s of targets

analyzed in a

single test, down to

the single

nucleotide

Turn around time

between 8-24

hours, and handle

upto 384 samples

per run

The solution works on

any current NGS

platforms.

All the steps are

optimized on liquid

handling systems to

enable minimal

hands-on time

Novel approach to

harness the power of

sequencing and keep

the costs competitive

PLUG AND PLAY

Urinary tract infections

Sexually transmitted infections

Tuberculosis

Post-operative infections

Meningitis

Infectious Diseases

Applications

In 2018

10,400,000 cases

1,600,000 deaths

~2 billion people carry the disease

Increasing drug resistance

480,000 cases

230,000 deaths

Drug resistant tuberculosis – a global challenge

Rifampicin

Isoniazid

Pyrazinamide

Streptomycin

Ethionamide

Kanamycin

Capreomycin

Amikacin

Rifabutin

Prothionamide

Levofloxicin

Gatifloxicin

Ciprofloxicin

Ofloxicin

Moxifloxicin

PAS Cycloserine

Delamanid

Clofazamide

Bedaquiline

Carbapenems

PA-824

Linezoid

β-lactams

SQ-109

BTZ

Trimethoprim

Macrolide

NAS-21/NAS-91

M. tuberculosis

Lineage 2

Lineage 3

Lineage 4

Lineage 5

Lineage 6

Lineage7

Lineage 1

M. avium

M. bovis

M. africanum

M. canetti

M. caprae

Which tuberculosis?

Which antibiotic therapy?

Future antibiotics

Targeted DNA Sequencing

Machine learning

+

Aim: choose the correct initial therapy

Aim: choose the correct initial therapy

Advisors

• Algorithms• Financing• Leadership

CEO CSO BoardBoard

• Chemistry• Regulatory• R&D

• Infrastructure• Network• Supervision

• Network• Infrastructure• TB expert

100% 100% 10%10%

Dr. PrajwalDr. Sebastian

DümckeProf. Thorsten

BuchDr. Peter Keller

8 years bioinformatics1 year cancer

diagnostics

11 years molecular biology

Quality management

Role

Tasks

Experience

Time committed

10 years medical doctor6 years diagnostics

17 years genetics8 years leadership

Team

R&D

Business Development

Suppliers & distributors

Regulatory

BioEntrepreneurship

& Innovation

Fund raising

Market access &policy drivers

Partners