A program in design & validation · 2026

A parent with dementia comes home from the hospital. The family becomes the clinical team overnight.

LaterCare is a voice companion for that month. It translates the discharge papers into plain language, calls every morning for 30 days, and tells the family and the doctor what it heard.

Every morning at 9:00, on a smartphone, a smart speaker, or the landline they already own.

Live demo · fictional patient
LaterCare
age 81 · pneumonia · Alzheimer's disease
0:00
Day 4 of 30
Incoming call · 9:00 AM

This is what Day 4 sounds like. About half a minute, then see exactly what his daughter receives.

What Anna receives, 9:04 AM

Morning check-in, Day 4. He needed a reminder of who we were, the same as Day 2 and Day 3, so nothing new there. His breathing is easier than yesterday, in his own words. He took his morning pills with breakfast. He was not sure whether he took the bedtime memory pill, so please check the pillbox when you get a chance. He asked about the dog twice. He would like you to call.

⚑ Bedtime dose uncertain · check pillbox ✓ Breathing improved ✓ Morning medications taken ◷ Orientation steady vs Day 2–3
0M
Americans age 65 and older living with Alzheimer's disease.
Alzheimer's Association · 2026
0%
Readmitted within 30 days when the patient has dementia, against 14.7% of matched patients without it.
JAMA Netw Open · 2023
~0M
Americans giving unpaid care to a family member or friend with dementia, 19 billion hours a year.
Alzheimer's Association · 2026
0%
Share of family caregivers who received any medical training for the care they give.
AARP / NAC · 2025
A short introduction

Why this exists, in the founder's own words.

Sumir Bassanpal · Founder Recorded for the families, clinicians, and funders deciding whether to spend time with us.
The paperwork, made usable

The discharge packet is written for clinicians. The person reading it is a daughter in a parking garage.

Anna photographs the packet during setup. LaterCare translates it line by line, at a fifth to sixth grade reading level, and every line links back to the original page. Nothing is invented.

From the discharge packet
DX: COMMUNITY-ACQUIRED PNEUMONIA (J18.9)HX: ALZHEIMER DISEASE, LATE ONSET (G30.1) AMOX-CLAV 875-125MG. 1 TAB PO BID x 7D.DONEPEZIL 10MG PO QHS. CONTINUE.INCENTIVE SPIROMETRY 10X Q1H WA.FALL PRECAUTIONS. AMBULATE W/ ASSIST.DIET: REGULAR. ENCOURAGE PO FLUIDS.O2 SAT >92% ON RA AT DISCHARGE.HOME HEALTH RN x2/WK x2 WKS. RTC IF FEVER >100.4F, ↑CONFUSION, SOB.F/U PCP 7D. F/U NEURO 30D.
What the family and the patient see
  • The infection medicine (amoxicillin), one tablet with breakfast and one with dinner, for seven days. Finish all of them, even after he feels better.
  • The memory medicine (donepezil), one tablet at bedtime. Nothing about this one has changed.
  • The breathing exercise, ten slow breaths every hour while he is awake. It keeps the bottom of the lungs open.
  • Call the doctor if he seems more confused than usual. New confusion is often the first sign an infection is coming back. We ask about it on every call, and we tell you what we heard.

Every line traces back to a line in the original paperwork, so a family or a clinician can always check the source. If the system cannot trace a statement, it does not make it.

Day 0
Discharge. Anna sets up in about ten minutes.
Day 1
First 9:00 AM call. Short and gentle.
Day 4 · today
The call you just heard.
Day 7
Checks the follow-up visit actually happened.
Day 14
Halfway summary to the family.
Day 30
Graduation note to the family and the doctor.
i.

Translates the paperwork into language a person can use.

Discharge instructions, medications, and follow-ups in clear, plain words. Every line links to its source page in the original packet.

English & Spanish at launch
ii.

Calls every morning, on whatever device is already in the house.

Smartphone, smart speaker, or the landline. The voice is built and tested for older speakers, including people who need re-orienting at the start of every single call.

About two minutes a day
iii.

Routes what it hears to the family and the doctor.

Plain-language notes to the family caregiver, structured notes to the primary care office. Every alert carries its reason and a link to what was actually said.

The doctor keeps clinical authority
The variable nobody has tested

Post-discharge programs have been tried for thirty years. The ones that work and the ones that don't look almost identical on paper.

We think the difference is not the technology. It is the question of who the system belongs to.

Most post-discharge programs are owned by the hospital. The hospital buys them, the hospital runs them, and they call the patient because the hospital needs a number to move. The family, who is doing the actual clinical work at home, is treated as a bystander. LaterCare inverts that architecture on purpose.

The family
purchases

The adult child already coordinating the care is the buyer. Not a health system procurement cycle that takes eighteen months and ends in a pilot.

The patient
owns

The account, the record, and the export belong to the older adult. They leave with all of it, whenever they want, in a format they can hand to anyone.

The caregiver
initiates

Setup, escalation preferences, and the decision to continue sit with the person doing the day-to-day work, not with a discharge coordinator they met once.

No trial has isolated that variable. Our feasibility study proves the system can be built and delivered to real families in real homes. The randomized trial that follows is the one that tests the architecture itself.

Where we are

A 40-dyad feasibility study, across 12 months.

Pre-launch, and honest about it. We have working components, not yet an integrated system validated with real patients. This is the study that closes that gap.

Design
Single arm

Every enrolled dyad receives LaterCare. There is no control group at this stage, by design.

Participants
40 dyads

A person living with dementia recovering at home after a hospital stay, together with their family caregiver.

Duration
30 days of calls

Daily morning check-ins for each dyad, inside a 12-month design and validation program.

Primary endpoints
Feasibility and usability

Can we recruit and retain dyads, deliver the calls reliably, and can families actually use the thing without help.

Secondary, exploratory
Caregiver burden

Measured with the Zarit Burden Interview short form (ZBI-12). Exploratory at this stage, not a claim.

What this study is not
Not a readmission trial

It is not powered to detect a change in readmissions. Claiming otherwise from 40 single-arm dyads would not be science.

Q1
Foundation

Caregiver interviews with dementia-screened dyads. Clinical and engineering leads on. Regulatory pathway confirmed.

Q2
Build

Translation pipeline complete. Voice tuned for older speakers and for re-orientation at the top of each call.

Q3
Test

Closed alpha in real homes. Safety and equity review. Caregiver distress protocol live before first enrollment.

Q4
Next phase

Randomized protocol finalized. Recruitment site partnerships in place for the trial that tests the architecture.

Rooted in evidence

Thirty years of transitional-care research points at one thing. When the family is built into the design, outcomes move.

Out of every 100 older adults who go home from the hospital, this is how many are back within 30 days. The only difference between these two groups is dementia.

21.5%Patients with Alzheimer's disease or a related dementia
14.7%Matched patients without dementia

Across 722,911 hospitalization episodes in a statewide collaborative, propensity-matched. That gap of roughly seven people in every hundred is the space LaterCare is built to work in. JAMA Netw Open · 2023

Educating the patient and the family cut readmissions.

A nurse-directed program that included comprehensive education for the patient and their family reduced heart failure readmissions by 56 percent in older adults. Three decades on, it is still the reference trial for this work, and it is the paper LaterCare is built on.

N Engl J Med · 1995

Caregiver integration at discharge reduces readmission.

A meta-analysis of 15 randomized trials covering 4,361 older patients found a 25 percent lower risk of readmission at 90 days when family caregivers were integrated into discharge planning.

J Am Geriatr Soc · 2017

Automated calling with a caregiver in the loop shows promise.

A randomized evaluation of an automated post-discharge calling program with caregiver feedback loops was directionally positive. It was underpowered rather than null, and caregiver integration was part of the design.

Jt Comm J Qual Patient Saf · 2020

Dementia raises the stakes on all of it.

Across 722,911 hospitalization episodes, 21.5 percent of patients with Alzheimer's disease and related dementias were readmitted within 30 days, against 14.7 percent of matched patients without dementia.

JAMA Netw Open · 2023

We are committed to honest reporting on what works and what does not, and to privacy a family can take with them. Clear sharing controls, and a portable export of everything created with us.

Who's building this

A founder with healthcare-system fluency, and senior clinical expertise now formally committed to the program.

Founder & Principal Investigator

Sumir Bassanpal, MBA

Sumir leads accounting operations inside one of the largest publicly financed healthcare programs in the United States. He brings federal-program financial fluency, healthcare operating experience, and the discipline to carry LaterCare through its design and validation phase.

He is candid about what he is not. He is not a clinician and he is not an AI engineer. Both of those are named, contracted, budgeted roles on this program rather than gaps papered over with confidence.

  • MBA, AACSB-accredited program · 4.0 GPA · Beta Gamma Sigma
  • BS, Accounting
  • Operating experience in publicly financed healthcare
Clinical consultants · committed

Two senior academic geriatricians

Both practice at a major American research medical school, and both have agreed to consult on this program.

One is a board-certified geriatrician who sees patients in a memory diagnostic center, has served as site principal investigator on Alzheimer's clinical trials, and is medical director of a CMS dementia care program.

The other is the first author of the 1995 New England Journal of Medicine trial described at the top of our evidence section, the study that established what family-inclusive post-discharge care can do.

They are named in full in our program documents. We will introduce them by name here once we have their written go-ahead to do so.
How it is built

Built for a regulated environment, from the first line of code.

HIPAA-aligned voice

Our voice platform was chosen specifically for its HIPAA posture and its business associate agreement, over faster alternatives that could not offer one.

Traceable, not generative

Every plain-language line maps to a line in the discharge packet. The system does not invent clinical content, and it does not guess.

Built to talk to the record

SMART-on-FHIR interoperability, so summaries can reach the systems clinicians already use instead of becoming one more portal to check.

The family keeps the data

Clear sharing controls and a portable export of everything created with us. If a family leaves, they leave with their record intact.

Three ways to engage

We're talking to three audiences. Tell us which one is you.

For families

Caring for a parent with dementia who has been in the hospital?

We are not live yet. Join the waitlist and we will write when the first cohort opens. No spam, and your information stays with us.

For clinicians & researchers

Work in dementia care, transitions of care, or caregiving research?

We would be glad to talk. LaterCare is a scaffold that sits alongside the transitional-care model you already run, and our feasibility study is being designed with senior clinical input.

Open a conversation
For funders & advisors

Funder, foundation, or prospective advisor?

We share the full program proposal under a confidentiality agreement, structured around clear, measurable artifacts and dated milestones rather than promises.

Request the proposal