TruScreen Invited to Present WHO AI Collaboration Meeting

NZX/ASX Announcement
13 January 2025
TruScreen invited to present at WHO global AI collaboration
meeting
• TruScreen presented to a World Health Organisation (WHO) global meeting to further the
use of AI technologies for screening of cervical cancer
• The presentation by CEO, Marty Dillon is attached for the information of stakeholders
At the invitation of World Health Organisation (WHO) TruScreen Group Limited (NZX/ASX:TRU)
presented at a WHO global AI collaboration meeting on 11 and 12 November 2024 in Edinburgh. The
meeting investigated the further use of Artificial Intelligence (AI) technologies for cervical cancer
screening.
TruScreen’s unique AI enabled technology was the only opto-electric tissue differentiating medical
device company invited to participate.
This invitation followed on from the UNITAID Report that featured TruScreen as a technology, currently
in commercial use, for the primary screening of cervical cancer. After years of clinical trials, TruScreen
technology received recent recognition by peak Medical Organisations and national Government
agencies including:
• the Chinese Obstetricians and Gynaecologists Association (COGA),
• the Chinese Society for Colposcopy and Cervical Pathology (CSCCP),
• the Vietnam Ministry of Health National Technical List,
• COFEPRIS, the Mexican public health regulator, and
• The Russia Cervical Cancer Screening Guidelines.
TruScreen Chair, Mr Tony Ho commented:
“This WHO invitation was significant for TruScreen, and signals that, along with the recent recognition by
national government agencies and peak Medical Organisations, that WHO recognises the use of
TruScreen’s unique AI enabled opto-electric technology to reduce the preventable deaths of women
from cervical cancer.
This is particularly relevant in TruScreen’s target markets – countries with limited or no cervical cancer
screening programs, that resulted in high cervical cancer mortality rates. Ninety percent (90%) of
cervical cancer deaths worldwide occur in these countries.”

This announcement has been approved by the Board.

Ends

For more information, visit

www.truscreen.com or contact:
Martin Dillon
Chief Executive Officer
martindillon
@truscreen.com
Guy Robertson
Chief Financial Officer
guyrobertson@truscreen.com


About TruScreen:

TruScreen Group Limited (NZX/ASX: TRU) is a medical device company that has developed and
manufactures an AI-enabled device for detecting abnormalities in the cervical tissue in real-time via
measurements of the low level of optical and electrical stimuli.
TruScreen’s cervical screening technology enables cervical screening, negating sampling and
processing of biological tissues, failed samples, missed follow-up, discomfort, and the need for costly,
specialised personnel and supporting laboratory infrastructure.
The TruScreen device, TruScreen Ultra
®
, is registered as a primary screening device for cervical cancer
screening.
The device is CE Marked/EC certified, ISO 13485 compliant and is registered for clinical use with the TGA
(Australia), MHRA (UK), NMPA (China), SFDA (Saudi Arabia), Roszdravnadzor (Russia), and COFEPRIS
(Mexico). It has Ministry of Health approval for use in Vietnam, Israel, Ukraine, and the Philippines,
among others and has distributors in 29 countries. In 2021, TruScreen established a manufacturing
facility in China for devices marketed and sold in China.
TruScreen technology has been recognised in CSCCP’s (Chinese Society for Colposcopy and Cervical
Pathology) China Cervical Cancer Screening Management Guideline.
TruScreen has been recognised in a China Blue Paper “Cervical Cancer Three Stage Standardized
Prevent and Treatment” published on 28 April 2023.
In Dec 2023 TruScreen technology was added to the Vietnam Ministry of Health approved National
Technical List, for use in Vietnam’s public and private healthcare sectors
In financial year 2024 alone, over 200,000* examinations have been performed with TruScreen device.
To date, over 200 devices have been installed and used in China, Vietnam, Mexico, Zimbabwe, Russia,
and Saudi Arabia. TruScreen’s vision is “A world without the cervical cancer”
©
.
To learn more, please visit: www.truscreen.com/.
*Based on Single Use Sensor sales.

Glossary:
Pap smear (the Papanicolaou smear) test involves gathering a sample of cells from the
cervix, with a special brush. The sample is placed on a glass slide or in a bottle
containing a solution to preserve the cells. Then it is sent to a laboratory for a pathologist to examine
under a microscope. https://www.cancer.net/navigating-cancer-care/diagnosing-cancer/tests-and-
procedures/pap-test
LBC (the liquid-based cytology) test, transfers a thin layer of cells, collected with a brush from the cervix,
onto a slide after removing blood or mucus from the sample. The sample is preserved so other tests can
be done at the same time, such as the human papillomavirus (HPV)
test https://www.cancer.net/cancer-types/cervical-cancer/diagnosis

HPV (human papilloma virus) test is done on a sample of cells removed from the cervix, the same
sample used for the Pap test or LBC. This sample is tested for the strains of HPV most commonly linked
to cervical cancer. HPV testing may be done by itself or combined with a Pap test and/or LBC. This test
may also be done on a sample of cells which a person can collect on their own.
https://www.cancer.net/cancer-types/cervical-cancer/screening-and-prevention


Sensitivity and specificity mathematically describe the accuracy of a test which reports the presence
or absence of a condition. If individuals who have the condition are considered "positive" and those who
don't are considered "negative", then sensitivity is a measure of how well a test can identify true positives
and specificity is a measure of how well a test can identify true negatives:
• Sensitivity (true positive rate) is the probability of a positive test result, conditioned on the
individual truly being positive.
• Specificity (true negative rate) is the probability of a negative test result, conditioned on the
individual truly being negative (Sensitivity and specificity – Wikipedia).
For more information about the cervical cancer and cervical cancer screening in New Zealand and
Australia, please see useful links:
New Zealand: National Cervical Screening Programme | National Screening Unit (nsu.govt.nz)

Australia: Cervical cancer | Causes, Symptoms & Treatments | Cancer Council

---

TruScreen
Reproducibility of Test Result &
Test-Retest Reproducibility
11 November, 2024
1

About
TruScreen
Technology
2

3
TruScreen
Primary cervical cancer screening device for
detection of pre-cancerous and cancerous
cervical tissue via an AI generated Algorithm
Handheld device
What is the TRU System and how does it
work? The TRU system consists of a
handheld device (HHD), intelligent cradle and
a single-use-sensor (SUS).
Intelligent Cradle
Single Use
Sensor (SUS)

4
TruScreen Regulatory Approvals
International Quality
Accreditation:
•ISO 13485
•ISO 60601-1-2
•CE Mark
International Approvals:
•CE Mark, European Union
•NMPA, China
•TGA, Australia
•MHRA, UK
•SFDA, Saudi Arabia
•Roszdravnadzor, Russia
•COFEPRIS, Mexico
•WAND New Zealand
•Zimbabwe Ministry of Health
•IEAKI Indonesia
•Vietnam Ministry of Health

5
Optical Tissue Differentiation
TruScreen measures the
scattering and diffuse reflection
of Distant Red, Red, Infrared and
Green light.
TruScreen detects changes in
sub-surface tissue that are not
visible in visual inspection or
collected in a Pap Smear sample.

6
Electrical Tissue Differentiation
Squamous tissue acts as a
battery and stores, for a brief
period, electrical charge.
TruScreen stimulates the cervix
with low voltage multi pulse
stimulation (0.78 V) and then
measures the voltage decay of
the tissue.

7
Algorithm and a Repeatable Result

8
TruScreen Algorithm
Developed by PLT/CSIRO / University of Sydney
•The Algorithm Team were led by Geoff Mckellar and Stephen Gould at PLT and David McMichael at the
CSIRO, and a PLT team led by Victor Skladnev developed the ‘probe’ and signal processing technology.
•Algorithm development utilised mathematical techniques including PCA, mixture models,
clustering/vector quantization, SVM, neural network, logistic classifiers.
•Mixture models and logistic classifier gave the best performance

9
TruScreen Algorithm
•The Algorithm D2.03G was then ‘frozen’ and provided that the TruScreen Device is in ‘spec’ and the
users follow the IFU then a reproducible/repeatable result is assured
• This has been clinically verified in trials involving more than 40,000 women, in multiple settings and
across multiple ethnicities
Improvements using
o Feature engineering (e.g. add Fourier transform based features)
o Support Vector Machine
o Random Forest
o Sparse expectation–maximization (EM) algorithm
o Monte Carlo method
Showed no improvement on the TEST database, even though they showed improvement on the
development database.

10
Reproducibility of TRU Result
The TruScreen Algorithm is fixed and processes data using the same ‘cluster’ definitions
for every patient:
But:
Equipment and People vary thus the control of the quality of input date is essential:
Equipment Variability to be stabilised:
•Handpiece parameters
•SUS Parameters
•Ageing effect on both
•Start UP Self Check – Electrical and Optical
•OTP Test
•SUS Fit test – Electrical
•SCS at 20 Tests (Gain/Drive Current)
•Probing Pattern
Human Variability to be stabilised
•Follow the IFU
oContraindications
oPatient Preparation
oProbing Pattern
•TRAINING
Reproducible Results Require Reproducible Equipment and Reproducible Users
CREATE a DEVICE AGNOSTIC SOLUTION

11
Reproducibility of an Algorithm
The use of AI to enable an algorithm to ‘self-improve’
raises many questions.
•If an Algorithm is constantly ‘improving’ how is it clinically
verified and validated.
•If the Algorithm is ‘improved’ via input data, how is the
input equipment controlled so that the input data is
constant, and not compromised by variable background
‘noise’
•Light Source, cameras etc. As these age and the intensity
and colour of the background light changes, how does the
equipment compensate for these variations?
Self Learning Algorithms are
meant to be Self Improving
BUT
If Poor Data enters the
learning process will the
Algorithm learn ‘bad habits’
And
Become less accurate rather
than more accurate.

12
Reproducibility of an Algorithm
•As cameras change, how does the algorithm compensate or
adjust to the differing optics of the new camera?
•Solarscan lessons....perfect colour and prefect light for
melanoma screening
•If the Algorithm is ‘improved’ via input data, how is the
human element controlled so that the input data is constant,
and not compromised -
oUser training and technician training

13
What is a Gold Standard
If the AI inputs are measured against a database of
classifications derived from ‘Gold Standard’ diagnoses,
how Gold is that Standard?
Subjective analyses of images, laboratory handling failures,
poor sample collection and processing can all devalue the
Gold in any standard.

14
What is a Gold Standard
•Colposcopy
•Histology
•HPV DNA
•All have human input.....
•Lessons from around the world show that a ‘suspicious’
mind will guard against blind trust
Thus Reproducible Results Require not just Reproducible
Equipment and Reproducible Users, but an unvarying Gold
Standard