Training Data Collection & AI Development (LC_SSP)

The aim of this study is to facilitate collection of real-world data to test and train the analytics engine for each prototype algorithm. Preliminary datasets will be generated to enable a dry run of the prototype algorithms to check their predictive functionality as part of simulated 'experimental' scenarios at each LifeChamps partner site. This preparatory work will be critical to the development of the LifeChamps platform, prior to progressing to a larger scale feasibility trial.

"The LifeChamps project (https://lifechamps.eu/) is creating a digital platform to support clinical teams to provide more integrated follow-up care to older patients with cancer. The digital platform will integrate data coming directly from the patient (patient-reported outcomes and sensor data from wearable devices), from the home environment (home sensors, weight scales), and from the clinical environment (data routinely collected via the Electronic Health Record). The digital platform will use big data analytics (machine learning) to process all data as part of predictive clinical algorithms for frailty and quality of life for older patients with cancer. Development of each clinical algorithm requires that the prototype model (or analytics engine) is trained using abundant real-world data to help consolidate the predictive ability and validity of the algorithms before the algorithms are deployed in a larger scale feasibility trial. A prospective, time series design will be employed, whereby the LifeChamps platform will be deployed for a total of 3 months. Older patients with a cancer diagnosis will be the target population for this study. Consecutive sampling will be used, whereby all older patients with cancer who meet the eligibility criteria will be approached and invited in the study. Each study participant will be involved in the study for 3 months in total. A 3-month recruitment period will be allowed, bringing the total study duration to 6 months (from first patient being enrolled until last patient finishing data collection). AUTH: Patients aged 65 years and above, diagnosed with early-stage breast or prostate cancer will be identified at the Department of Medical Oncology at G. Genimatas General Hospital, "Alma Zois" a non-profit association of women who had experienced breast cancer based in Thessaloniki, Greece and collaborating private clinics. The patients will be presented with the opportunity to participate in the study and screened based on the inclusion and exclusion criteria. Potential participants will be provided with the information sheet and the consent form, informed that should they decline to participate this will not change their current treatment and provided the opportunity to ask any questions they may have. Region Stockholm : Melanoma survivors aged 65 years and above, who have completed primary treatment within the last 12 months prior to the study, will be identified by our clinical partners within primary and secondary care in Region Stockholm. Online advertisements for the study will also be disseminated and participants may be recruited through LifeChamps partners (e.g. Karolinska Institute) or the Swedish melanoma patient association (Melanomföreningen) which will direct them to an online recruitment form within Region Stockholm. The researchers will then make a follow-up call to confirm interest in study participation. During the contact, the melanoma survivor will be asked to provide their written consent or decline participation and ask for a second contact before his/her final decision. All declining melanoma survivors will be thanked for their time and reassured that their decision will have no impact on their current or future treatment and care. All consenting participants will be reassured that they can withdraw at any time that they desire from the study. After written informed consent has been provided, the mini-COG will be used to evaluate study participants' cognitive function and impairment at baseline. The mini-COG consists of a 3-word recall and a clock-drawing test, and can be completed within 5 minutes. A score of less than 3/5 indicates the need to refer the patient for full cognitive assessment. The researcher will also arrange for study participants to receive study equipment, i.e. home sensors, wearable activity sensors, smart weight scale, and mobile app. The researcher will arrange a suitable time for a home visit to install the home sensors and test functionality. The researcher will demonstrate use of study equipment to the participant, and reiterate that support with use of the technology will be available. Data collection will involve a variety of sources, including the patient (patient-reported outcomes and sensor data from wearable devices), the home environment (home sensors, weight scales), and the clinical site (data routinely collected via the Electronic Health Record). The following technology will be used: Mobile devices: Activity tracker wristband (FitBit charge 4). It will be used to passively monitor and collect data on heart rate, heart rate variability, steps, activity tracking, sleep monitoring, breathing rate, skin temperature and SpO2. Mobile app (SALUMEDIA). It will be used to enable collection of patient-reported outcome measures (PROMs) and to forward this information along with the data gathered by the activity tracker and the smart scale to the Raspberry Pi Kit at home. At home sensors / devices: LOCS Home sensors: They will be used to monitor participants' daily activities e.g., to track ambulation and functioning. Study participants will be provided with 4 motion sensors, 1 door contact sensor, 2 corridor sensors, and a tag device. Smart Scale (Withings Body+): It will be used to measure weekly body weight, body composition and body mass index. Raspberry Pi (RPI) kit: As an edge gateway, RPI is hosting LOCS gateway, Movesense Gateway and data ingestion service. RPI will enable data collection and edge analytics and transfer of data to the LIFECHAMPS platform. Smart plug: It will be used to collect data about use home appliances and thereby data about active daily living. Selected study participant clinical and demographic data from the local EHRs will be collected and loaded onto the LifeChamps analytics engine. The data will be loaded by technical partners via the LifeChamps dashboard for processing and analysis. Data regarding recruitment rate (patients consenting / patients approached), participant retention in the study, reasons for study discontinuation (if offered), participant adherence with technology, issues with technology and need for troubleshooting will be recorded. These data will (a) be recorded by local researchers using bespoke 'recording logs' in the form of an Excel spreadsheet, and (b) remotely monitored and logged by technical partners involved in the distribution / management of the technology to be used in the trial as described above."

Single Group: Clinical trials with a single arm A prospective, time-series design will be employed, whereby the LifeChamps platform will be deployed for a total of 3 months.

Participants will be asked to use the LifeChamps platform and will be provided with the study equipment.

Participants will be provided with the study equipment, i.e., a mobile app, smartwatch, smart scale, location home sensor, a smart plug, and a micro-computer, with which they will need to interact with the devices for three months. Specifically, participants should wear the activity tracker wristband (Fitbit Charge 4) as much as possible. Additionally, participants should use the smart scale to weigh themselves and the mobile app to fill in selected PROMs monthly, while the ambient home sensors (location home senor (LOCS) and smart plug) will be passively collecting information about their everyday living during these three months. Lastly, participants' clinical and demographic data from the local EHRs will be collected.

The collected data will be used to train and test the analytical models. For example, the initial data from sensors will be refined, further, with statistical, spectral and supervised learning analyses to identify and extract possible patterns (e.g., activities of daily living) inside their signals. Sensor, EHR and PROM data will be all analysed together through exploratory algorithms (e.g., pairwise Markov random fields, Bayesian networks) to identify possible interactions and dependencies among their trajectories, mapping the frailty and QOL domains of elderly prostate and breast cancer patients across all the data collection process.

Core symptoms/signs of depression and anxiety as measured by PHQ-4 via 0-3 Likert scales, with 3 being the worst.

Initial assessment symptoms palliative care patients' symptoms as measured by ESAS-r via 0-10 Likert scales, with 10 being the worst, and a blank scale for patient-specific symptoms.

Level of adherence to medication as measured by MARS via a Medical adherence Report Scale (MARS) via "yes" or "no" questions attributed with a 0 or 1 point (according to the content of the question), which are then used to calculate a score of adherence.

Function-based risk assessment of health deterioration in older adults as measured by VES-13 via a No Difficulty-Unable to do Likert scale and Yes/No/Don't do questions. All these are attributed to specific points used to calculate the final score. Scoring ranges between 0 and 10, 10 being the worse outcome.

Cognitive and emotional representations of illness as measured by Brief Illness Perception Questionnaire via eight 0-10 Likert scaled questions and one open ended. The questionnaire does not have a best or worst outcome, rather it attempts to ascertain the patient's perception of the illness.

Tracking symptoms of prostate cancer as measured by International Prostate Symptom Score (IPSS) via 0-5 Likert scales, with 5 signifying higher frequency of symptoms.

Assessment of erectile/sexual function as measured by International Index Erectile Function-6 via 0-5 Likert scales, with 5, being the best outcome

This outcome will be calculated by dividing the number of participants consented over the number of approached participants. It is a ratio between 0 and 1, with 1 being the best outcome.

This outcome will be calculated by dividing the number of participants that finished the study over the number of participants that entered the study. It is a ratio between 0 and 1, with 1 being the best outcome.

This outcome will be calculated by dividing the number of times the technology was used over the maximum times that technology can be used over the course of the study. It is a ratio between 0 and 1, with 1 being the best outcome.

Inclusion Criteria: Breast or prostate cancer. Diagnosed with early stage (I-III) cancer (breast, prostate) and living beyond initial cancer treatment (curative/incurable). Diagnosed with advanced or metastatic disease with life expectancy >12 months. At least 1 month after a) local treatment with curative intent (surgery, radiotherapy) or b) initiation of systemic treatment (hormone treatment, CDK4/6 or new generation antiandrogens). Absence of diagnosed secondary malignancy. Deemed by a member of the multidisciplinary team as physically and psychologically fit to participate in the study. Able to read, write and understand the respective local language (greek). Achieve a score of above 2 on the Mini-Cog during the screening process. Able to bring and use own Android version 10 (or above) device during the study. Domestic 24/7 internet access via wi-fi and/or 4G mobile data (will be provided if unavailable). Exclusion Criteria: Currently receiving chemotherapy. Terminal cancer stage on palliative care. Survival prognosis of <18 months from the time of recruitment. Unwilling to provide written informed consent. Presence of internal medical device (e.g. pacemaker etc.)

Data collected during the Real-Life pilot study, will be made available for sharing via open access repositories, according to the Final Version of the Data Managements Plan of the project.