PsA digital phenotyping and inflammation drivers study

The level of disease activity in Psoriatic Arthritis (PsA) and the perception
thereof by the patients determines the actions the rheumatologist takes to
optimize treatment outcomes among patients with this disease. Currently,
disease activity is measured by a combination of clinical measures and
patients* self-reported symptoms and functional ability. This requires the
patients to visit the outpatient clinic at regular intervals, which during the
Covid-19 pandemic was not always possible. The use of questionnaires to collect
Patients* Reported Outcomes (PRO*s) is a feasible option for monitoring
patients at a distance. However, from a long term perspective, survey fatigue
is a known limiting factor of PRO*s. Currently, there is no valid alternative
for unobtrusive remote disease activity monitoring.
The widespread use of smart devices by the general population, such as
smartphones or smartwatches, provides opportunities to develop and study
possibilities for Unobtrusive Remote Disease activity monitoring (URD) using
behavioural data captured by the sensors embedded within the
smartphones/smartwatches. We hypothesize that a high level of disease activity
in PsA will lead to changes in physical activity as registered by a patient*s
smartphone and smart watch as compared to a low disease activity state. We also
hypothesize that the information acquired by digital biomarkers will be
comparable to the information received through clinical measures and PROs.
Additionally, digital biomarkers are likely to provide information on other
disease characteristics such as tiredness and sleep problems. Adding these will
enhance the discriminative ability of our approach. Last but not least, we
hypothesize that patients will see the use of smartphone data as a privacy-wise
safe and fair deal in return for better insight in their disease. And in low
disease activity the use of digital biomarkers could reduce the amount of
follow-up appointments at the out-patient clinic.

Primary objectives:
To develop and internally validate a novel and interpretable machine learning
model for detecting flare in PsA patients using integrated accelerometer data,
keystroke dynamics and screen time metrics (i.e., digital biomarker) to assess
changes in their physical activity patterns against clinical defined flare by
the rheumatologist. Accelerometer data is captured by both smartphone and
smartwatch.
To develop and internally validate machine learning models that capitalize on
sleep, fatigue, pain, stress, mechanical stress, composition of gut microbiome,
genetic risk and environmental exposure for flare prediction (either
clinically established or evaluated by the digital biomarker) in patients with
PsA.


Secondary objectives:
To assess construct validity of the novel and interpretable machine learning
model for detecting flare in PsA patients using integrated accelerometer data,
keystroke dynamics and screen time metrics (digital biomarker) to assess
changes in physical activity patterns against the continuous measure of
clinical composite scores of disease activity and impact of disease used by the
rheumatologist and impact of disease as reported by the patient
To develop and internally validate a novel and interpretable machine learning
model for changes in joint and skin appearance that relates to flare in PsA
patients using video analysis of hand, posture and gesture and photos of the
hands and feet against clinically defined flare by the rheumatologist
To determine intraperson reliability of the AI-driven digital biomarker system
To determine clinically relevant changes in the AI-driven digital biomarker
system
To determine minimal detectable difference in the AI-driven digital biomarker
system
To assess the intraperson variation of stress, mechanical stress and changes in
gut microbiome on the occurrence of flare
To identify genetic contribution to disease activity and pain
To evaluate costs and effects of the digital biomarker of future care to
current care
To evaluate the compliance and satisfaction of the users with the smartphone-
and smartwatch-based measurement of disease activity and flare.

One year international multicenter prospective observational cohort

Patients with PsA experience difficulties in dealing with unpredictable disease
activity which can have consequences on their daily living. With the
introduction of smart devices, they could have better understanding on the
disease influence on their physical activity patterns, stress levels, sleep,
pain, stiffness and fatigue. During the study, patients will benefit from the
features provided by the Garmin Smart watch Vivoactive 5 such as physical
activity and heart rate.
Digital monitoring of patients will be performed continuously for a year using
a smart watch that will be provided for patients to wear daily, and a data
capturing application (app) that will be installed on their smartphone. The
data capturing app will collect -unobtrusively- the keystroke dynamics and the
accelerometer/gyroscope data of the smartphone. The levels of pain, fatigue and
stiffness will be also monitored via questions provided by the app on these
symptoms. These are very short questions that appear one time a day for the
first 14 days, and are answerable within a few seconds. In addition, the app
allows the patients to self-register a disease flare by pressing the flare
button. Once patient registers a flare, these questions will appear again. When
flare button is off, patients are inquired every two weeks if they are flare
free. Besides, photos of hands and feet and videos of hands, posture and
gesture will be collected at baseline, every 6 weeks and when patient
experiences disease flare. Photos captured by the patient must contain only
hands and feet, otherwise they will be discarded. Regarding the videos of
hands, posture and gesture, only the time series of hand/body landmarks from
the raw videos will be saved (no raw videos will be stored).
Patients will be assessed at baseline and followed-up every 3 months for a
total duration of one year. Each study visit requires around 30 minutes.
Patients are clinically assessed for disease activity at study center and are
requested to fill questionnaires, at baseline and every 3 months. Saliva
collection for DNA analysis is performed at baseline only, using a salivary
collection kit at the study center or at home. For microbiome analysis,
patients are provided with a home kit to collect stool at baseline and when
experiencing disease flare for which they seek help from rheumatologist. In the
Netherlands, additional stool sample collection is requested from patients
attending centers that are participating in DEPAR at months 6 and 12. Depending
on the hospital facilities, patients can bring samples back during their
hospital visit or send it to a central location via postal mail. To analyse
cortisol from hair, 3cm of hair will be sampled at study center from posterior
vertex at baseline and every 3 months.
Burden on patients include more outpatient clinical visits than standard care,
response burden, continuous use of smartwatch and continuous monitoring of
patient activity for 12 months, taking photos and videos, providing hair and
stool samples, and providing saliva for DNA analysis.