The workflow before#
Elara Caring provides home-based care to more than 65,000 patients a day across 225 locations in 16 states. It grew through acquisitions, and each one brought its own EMR and EHR systems. When a referral came in, an intake specialist read the packet, often scanned, keyed patient details into separate systems, checked insurance one case at a time and priced the bid from a spreadsheet. A single referral took 4 to 6 hours of staff time, and moving a patient from referral to admission took 3 to 5 days. Manual entry held accuracy at 85 to 90%. Referral sources don't wait, and slow bids cost Elara patients it should have won.
What we built#
We built an automated intake workflow that does the reading, keying and first checks for Elara's intake team. Referral documents are read, classified and turned into clean patient records; insurance is checked; and each case gets a risk and margin estimate before anyone prices a bid. Intake specialists handle the exceptions and the judgment calls, so a bid can go out the same day the referral arrives.
Referral packets read automatically
Documents land through one intake path no matter which system sent them. OCR pulls text from scanned pages, and NLP models classify each document and extract patient details, diagnoses and treatment plans. The specialist opens a case that's already filled in.
Validation that routes only exceptions
Every record runs through automated validation, and only low-confidence fields go to an intake specialist. We put the human check at the field level instead of on every packet, so staff spend their time on the handful of entries that need a person, not on retyping the rest.
Risk and margin scored before the bid
Underwriting models score each patient's care complexity and estimate the margin on the case, and insurance eligibility checks run automatically. The team decides which cases to bid and how to price them with the numbers in front of them, instead of a spreadsheet built from memory.
A live view of the referral pipeline
Power BI dashboards show every referral by stage, flag cases that have stalled and track margin by service line. Intake managers see where cases are stuck today, and leadership sees which service lines earn their keep.
How we delivered it
We ran the work over about 16 weeks in four phases: environment setup and compliance validation, the data pipeline, model training and validation, and training with a phased rollout. The HIPAA-compliant environment was designed with Elara's security team from the first week. We worked with IT, operations and clinical staff throughout, and rolled the workflow out location by location so intake teams could learn it on live referrals without a cutover day.
Before and after#
| Measure | Before | After |
|---|---|---|
| Referral to admission | 3 to 5 days | Under 24 hours |
| Staff time per referral | 4 to 6 hours | Under 45 minutes |
| Bid response time | 2 to 3 days | Same day |
| Intake data accuracy | 85 to 90% | 98% |
| Insurance eligibility | Checked by hand, one case at a time | Checked automatically at intake |
What's next#
The intake workflow now feeds clean, structured patient data into one warehouse. The next step is applying the same pattern to claims and denial management, where specialists still review most cases by hand and the same read, validate and route approach can cut the queue.
