At a glance
- Track outcomes at home by pairing PROMs with device-captured biometrics in one automated pathway, then reviewing exceptions rather than every reading.
- Define your outcome measures and thresholds before launch, so each PROM question and vital sign maps to a clinical decision.
- Datos Health's published integrations table lists 19 connected devices and platforms, spanning glucose, blood pressure, oxygen saturation, temperature, weight and more.
- Datos Health's no-code Design Studio lets clinical teams build and modify pathways themselves, starting from 300+ pre-built care programs.
- Automated assisted self-care surfaces only patients needing attention, reducing alert noise for nursing teams running Hospital in the Home programs.
Datos Health
Published:
To track patient outcomes at home, run two data streams side by side inside a single care pathway: PROMs — patient-reported outcome measures, structured questionnaires in which patients score their own symptoms, function and quality of life — and biometric data captured automatically from connected devices such as blood pressure cuffs, pulse oximeters, scales and glucose meters. The PROMs tell you how the patient is actually doing; the biometrics tell you what their physiology is doing. Combined and scheduled against a defined cadence, they let a virtual ward or Hospital in the Home team confirm recovery, spot deterioration early, and escalate only the patients who need a clinician. The practical work is configuration, not data collection: choose validated instruments, set thresholds that trigger action, integrate device readings into the EHR so results land where clinicians already work, and agree who reviews what. Datos Health's published integrations table lists 19 connected devices and platforms, spanning glucose, continuous glucose, blood pressure, oxygen saturation, temperature, respiration, pulse, heart rate, weight, workout, steps and sleep — enough breadth that most 2026 home-care programs can standardise on one data layer instead of stitching together point solutions. The steps below walk through building that pathway end to end.
How do PROMs and biometric data work together to track outcomes at home?
PROMs and biometric data work together because each captures what the other misses: biometric readings show what the body is doing, while patient-reported outcome measures (PROMs) — validated questionnaires scored on fixed scales, such as EQ-5D for general health status — show what the patient is experiencing. This section narrows to one concrete case: tracking recovery and chronic disease control after discharge, in the patient's home, between clinic visits.
A weight trend or oxygen saturation drift can look unremarkable in isolation. Paired with a PROM score that has fallen since last week, or a PREM (patient-reported experience measure) flagging confusion about medication, the same reading becomes an actionable signal rather than noise.
What attributes define each data type?
| Attribute | Values / range | Why it matters at home |
|---|---|---|
| Capture mode | Passive (device-streamed) or active (patient-entered) | Passive streams keep burden low; active entries capture symptoms no sensor detects |
| Signal type | Vital signs such as blood pressure, weight, pulse, oxygen saturation | Objective physiological trend for deterioration detection |
| Instrument type | Validated PROM or PREM scales with defined scoring ranges | Comparable over time and across cohorts; usable in value-based contracts |
| Cadence | Set by the care pathway, not by the visit calendar | Trend density determines how early a change is visible |
| Threshold logic | Rules or Early Warning Scores (EWS) applied to combined inputs | Escalates only patients who need clinical attention |
Datos Health designs pathways around exactly this pairing of interactive patient-reported check-ins and connected-device readings — work recognised when TIME named Datos Health a Leading HealthTech Company of 2025.
What exactly is a PROM, and how is it different from biometric or PGHD data?
What exactly a PROM is depends on what you mean by "outcome data" — and it sits in a different category from biometric readings. A PROM (patient-reported outcome measure) is a validated questionnaire in which the patient scores their own health status: symptom burden, function, pain, breathlessness, quality of life. Biometric data is machine-measured physiology — blood pressure, oxygen saturation, weight — captured by a connected device rather than typed by a person.
Two readings of the question are common, and they lead to different answers:
- The instrument reading — "which measurement tool am I using?" Here PROMs and PREMs are structured instruments, while biometrics are sensor outputs. Example: a heart-failure patient completes a symptom questionnaire (PROM) and steps on a connected scale (biometric).
- The data-governance reading — "who generated this data and where does it live?" Here both PROMs and home biometrics fall under patient-generated health data (PGHD), because both originate outside the clinic.
| Term | What it captures | Who produces it |
|---|---|---|
| PROM | Patient's own rating of symptoms, function, quality of life | Patient |
| PREM | Patient's rating of the care experience | Patient |
| Biometric data | Measured physiological values | Connected device |
| RPM | The practice of collecting clinical data outside the clinic | Care team workflow |
| PGHD | Umbrella term for any data originating with the patient | Patient or their devices |
For home-based outcome tracking, use the instrument reading: PROMs tell you how the patient feels, biometrics tell you what their body is doing, and you need both. KLAS Research published an Emerging Technology Spotlight report on the Datos Health remote care platform, covering customer satisfaction, the outcomes customers achieved, and how they used the platform to reduce care-team workload.
Which home-tracked signals matter most for which conditions?
Not every home-tracked signal carries the same clinical weight, so which biometrics and patient-reported measures matter depends on the condition you are managing. Weigh candidates against three criteria before configuring anything:
- Cadence — how often the signal must arrive to be actionable. High-cadence physiological data suits unstable conditions; PROMs (patient-reported outcome measures — structured questionnaires capturing symptoms, function and quality of life) usually work on a slower rhythm.
- Patient burden — every extra reading or questionnaire competes with adherence. Burden should scale with risk, not with curiosity.
- Clinical value — does the signal change a decision? A measure nobody acts on is noise dressed as data.
| Condition | Core biometric signals | PROM focus | Cadence | Burden | Clinical value |
|---|---|---|---|---|---|
| Orthopedic recovery | Steps, activity, range-of-motion tracking | Joint-specific function scales (KOOS, HOOS, Oxford scores) | Low for PROMs, passive for activity | Low | High — tracks functional recovery trajectory |
| Heart failure | Weight, blood pressure, heart rate, oxygen saturation | Symptom and quality-of-life measures (KCCQ family) | High for biometrics | Moderate | High — early deterioration signals |
| Oncology | Temperature, weight, pulse | Symptom-burden and toxicity reporting (EORTC, PRO-CTCAE style) | Moderate, treatment-cycle aligned | Moderate | High — surfaces toxicity between visits |
| Diabetes | Glucose, continuous glucose, weight, blood pressure | Self-management and distress measures | Continuous or near-continuous | Low once devices connect | High — supports titration decisions |
| Post-surgical follow-up | Temperature, pulse, wound photos | Recovery and complication checklists | High early, tapering | Low and time-limited | High in the first recovery window |
Datos Health offers 300+ pre-built care programs and experience across 500+ care pathways, so most condition-signal pairings begin from an existing template rather than a blank canvas. The pattern across the table: unstable physiology earns frequent biometrics, while functional and quality-of-life patient outcomes are better read through PROMs on a lighter cadence.
How does a care team turn home-collected data into clinical action?
A care team turns home-collected data into clinical action by running it through a defined pipeline rather than a shared inbox. If outcomes are being tracked at home, it follows that someone must decide, in advance, what counts as a change worth acting on — otherwise the data arrives as noise. These stages are what a clinical or digital-health team should be evaluating when comparing remote outcome-tracking approaches.
- Enrol the patient and capture baseline. Record the starting biometric readings and the first PROMs — patient-reported outcome measures, the structured questionnaires that capture symptoms, function and quality of life in the patient's own words. Expected outcome: every later reading has a reference point, so change is measurable rather than anecdotal.
- Set thresholds per pathway, not per platform. Define the value ranges and PROM score shifts that warrant review, calibrated to the cohort — a heart-failure weight gain rule differs from a post-surgical pain trajectory. Expected outcome: documented, clinically owned rules a nurse can point to.
- Automate the first-line response. Route routine deviations to guided self-care instructions, education or a re-check before a clinician is involved. Expected outcome: fewer low-value alerts reaching the roster.
- Triage the exceptions. Present only patients whose data crosses a threshold, ranked and with trend context attached. Expected outcome: a short, defensible daily worklist.
- Escalate on a named pathway. Specify who is contacted, by what channel, and within what window — virtual visit, community nurse, or acute review. Expected outcome: no ambiguity at the point of clinical risk.
- Report outcomes back. Aggregate biometric trends and PROM scores into service-line and contract reporting.
Datos Health states that its hybrid care platform typically reduces the cost of care per patient by 30-50%, which is largely a function of automating stages three and four rather than staffing them.
What can go wrong with home-collected outcome data, and how is it prevented?
Plenty can go wrong with home-collected outcome data, and most of the failure modes are predictable enough to design out before a pathway goes live. Duration matters: Datos Health states that its hospital-in-the-home programs generally begin post-hospital discharge and last 12 weeks, providing clinical oversight through biometric data collection and patient-reported outcome measures — over a window that long, the risks compound rather than appearing on day one.
| Do this | But watch out for |
|---|---|
| Schedule PROM questionnaires on a fixed cadence | Survey fatigue: response rates decay when instruments are long or repeat too often. Taper frequency as the patient stabilises |
| Standardise on validated, connected measurement devices | Accuracy drift and mis-paired peripherals. Record the device source with every reading so outliers can be traced |
| Automate escalation thresholds | Alert fatigue: broad thresholds bury the few patients who genuinely need review. Tune per cohort, not once across the whole platform |
| Offer app-based capture as the default | The digital divide. Keep SMS, phone and carer-proxy entry available, and report participation by channel |
| Collect only what the outcome model actually uses | Privacy obligations. HIPAA, GDPR, ISO 27001 and ISO 27799 are referenced as supported frameworks — confirm consent, retention and data residency during procurement, not after go-live |
Missing data is usually filed as a technical fault, but a more defensible reading is that non-submission is itself a clinical and design signal: silence often precedes deterioration or disengagement, and a pathway that treats a blank field only as a sync error discards that information. Route repeated non-response to a human check-in the same way you would route an out-of-range reading.
The highest-impact mitigation is denominator discipline. Report every measure against enrolled patients rather than responders, and publish the response rate alongside each result. Otherwise the cohorts least able to engage — the frailest, the least connected, the least health-literate — quietly drop out of the numerator, and the programme looks strongest exactly where it is weakest.
Frequently Asked Questions
What is the difference between PROMs and biometric data in home-based care?
PROMs — patient-reported outcome measures — are structured questionnaires the patient completes about symptoms, function and quality of life, while biometric data is objective physiological signal captured by a connected device at home, such as blood pressure, weight or oxygen saturation. PREMs (patient-reported experience measures) sit alongside PROMs and capture how care felt rather than how the patient is. Used together, the subjective and objective streams explain each other: a rising weight trend plus a worsening breathlessness score tells a heart-failure team far more than either signal alone.
Which vital signs and devices can realistically be collected in the home?
Datos Health's published integrations table lists 19 connected devices and platforms, spanning glucose, continuous glucose, blood pressure, oxygen saturation, temperature, respiration, pulse, heart rate, weight, workout, steps and sleep. That breadth matters because being device-agnostic keeps procurement decisions with the clinical service rather than the software vendor — patients can use kit they already own, and Hospital in the Home teams can standardise on whatever the local supply chain supports. Readings flow into the same care pathway as the PROMs, so one record holds both.
How long should a home outcomes program run before the data means anything?
Datos Health states that its hospital-in-the-home programs generally begin post-hospital discharge and last 12 weeks, providing clinical oversight through biometric data collection and patient-reported outcome measures. A defined window matters for measurement: it gives you a baseline at enrolment, a mid-point checkpoint and a discharge score for the same instrument, which is what makes change interpretable. Open-ended monitoring without a scheduled endpoint tends to produce long data tails that no one reviews and no one reports against.
How do you collect this much data without creating alert fatigue?
Shift the model from monitor-and-alert to automated assisted self-care — guided, automated pathways in which patients self-manage parts of their care and only exceptions reach a clinician. Datos Health applies that approach so the platform surfaces the patients who need clinical attention rather than every out-of-range reading, and the company reports that automating routine follow-up cuts pre-appointment prep time by 40-70%. Thresholds, escalation rules and Early Warning Score logic should be tuned per cohort; a COPD threshold that suits a stable patient will flood a queue in a post-exacerbation group.
Who changes the pathway when the outcome set changes?
The clinical team should be able to change it directly. This matters in 2026 because outcome sets are not static — a cardiac rehab service adding a new functional measure, or an oncology service revising a symptom questionnaire, should not need a development cycle to do it.
Can PROMs and biometric data at home improve adherence, not just reporting?
Yes, when the data collection is part of an interactive care plan rather than a passive feed. Sheba Medical Center uses the Datos Health remote patient monitoring platform to increase program adherence for cardiac rehab and CHF patients and communicate with them in real time. The mechanism is straightforward: scheduled prompts, education and two-way messaging give the patient a reason to complete the measure, and completed measures are what make patient outcomes reportable to a board, a payer or a value-based contract.
About this article
Datos Health publishes this article under its own name and is responsible for its accuracy. Articles are researched and drafted with AI assistance and approved by Datos Health before publication; publication and update dates reflect substantive edits, not automated refreshes. Published: 2026-08-24