Healthcare analytics is not a single capability. It is a maturity ladder — and where an agency sits on that ladder determines what their data can actually do for clinical decision-making, performance management, and patient outcomes.
The four types of analytics in healthcare are descriptive, diagnostic, predictive, and prescriptive. Each represents a different level of analytical sophistication, a different question it answers, and a different degree of clinical value. Most home health and hospice agencies are operating primarily in the first two tiers. The difference between where most agencies are and where the highest-performing agencies are competing is the move to predictive.
Descriptive Analytics: What Happened
Descriptive analytics summarizes historical data to show what occurred across a patient population, an episode, or a reporting period. It answers one question: what happened?
In home health and hospice, descriptive analytics is the foundation of most current reporting infrastructure: star rating summaries, OASIS completion rates, visit counts, episode discharge outcomes, hospitalization tallies by quarter. These reports are accurate and useful for retrospective review. They tell leadership what the agency's performance looked like last month or last quarter.
The limitation is structural. Descriptive analytics tells you what already occurred. It cannot tell you what is happening right now across your active patient census, and it cannot tell you what is likely to happen before the next clinical event occurs. For agencies managing PDGM performance and value-based care contracts, that limitation matters.
Diagnostic Analytics: Why It Happened
Diagnostic analytics goes one step further by identifying contributing factors and correlations in historical data. It answers the question: why did this happen?
A diagnostic tool might reveal that a cluster of hospitalizations in the previous quarter was concentrated among patients with a specific diagnosis mix and a high number of missed skilled visits in the second week of their episodes. That correlation is useful for retrospective quality review and targeted process improvement. It identifies patterns after the fact that can inform future care protocols.
Diagnostic analytics is more actionable than descriptive reporting, but it still operates in the past. The patients whose data generated the insight have already been discharged. The episodes have already closed. The outcomes have already been determined.
Predictive Analytics: What Is Likely to Happen
Predictive analytics is where the clinical value of data makes its largest leap. Rather than analyzing what happened or why, predictive analytics uses machine learning models trained on historical patient populations to generate forward-looking probability estimates: which patients are most likely to be hospitalized, which are trending toward a PDGM benchmark miss, which are approaching a clinical transition that warrants immediate attention.
For a home health clinical supervisor managing forty active patients, the practical difference is this: instead of reviewing yesterday's outcomes, the supervisor opens a census-level view each morning that shows which patients carry elevated risk today — before anything has gone wrong. The intervention window exists. The clinical decision about what to do with it belongs to the clinician.
This tier requires something the first two do not: a predictive model trained on enough of the right clinical data to generate accurate risk estimates. In home health and hospice, that means models trained specifically on post-acute patient populations — not general healthcare data applied to a care setting it does not understand.
Prescriptive Analytics: What Should Be Done
Prescriptive analytics extends the predictive tier by recommending specific actions in response to predicted outcomes. It answers: given what is likely to happen, what should we do?
In healthcare, prescriptive analytics is still maturing. Clinical care involves variables — patient preference, physician judgment, family context, care setting constraints — that make automated recommendations more complex than in other industries. The most credible applications currently sit close to the predictive tier: risk-stratified care pathways, resource allocation recommendations based on census-level risk distribution, and intervention prompts calibrated to a patient's specific clinical trajectory.
For home health and hospice agencies, the practical priority is not prescriptive capability — it is closing the gap between descriptive reporting and predictive analytics. That move alone represents a fundamental shift in how clinical risk is managed and how proactively agencies can respond before outcomes are determined.
Explore how Mosai's predictive analytics moves home health and hospice agencies up the analytics maturity ladder: mosai.com/platform/clinical-management