# Healthcare

> Connect patient, facility, drug and referral data across clinics, hospitals and ministries, so clinicians and planners can see capacity, stock and outcomes without more paperwork. Vertical ontology, workflows and compliance notes for healthcare in Africa.

Canonical URL: https://www.boltcliq.com/industries/healthcare
Part of: https://www.boltcliq.com/llms.txt

Connect patient, facility, drug and referral data across clinics, hospitals and ministries, so clinicians and planners can see capacity, stock and outcomes without more paperwork.
- Sector pack: Starting ontology for facilities, patients, stock and referrals
- EMR + registers: Typical sources: facility records, paper registers, lab and pharmacy
- Health data laws: Regulators in scope: data protection and health ministries
- SMS to tablet: Channels: SMS, WhatsApp, tablets and offline apps
### The challenge
- Health data in most African systems is spread across facilities that each run on their own. Some clinics use an electronic medical record. Many use paper registers, with monthly counts typed into a national reporting system. Laboratories, pharmacies and imaging run separate tools. Drug supply chains report stock on spreadsheets or in a logistics system that clinicians never see. Referrals between a health post, a district hospital and a specialist center often depend on a phone call and a handwritten note.
- Because records do not follow the patient, clinicians repeat tests and miss history. Stockouts of essential medicines are found after patients have been sent home. District managers see last month's numbers, not this week's. Planners cannot tell which facilities have the staff, beds and equipment to take a referral. Any AI project needs consented, labeled, secure data, and most facilities cannot provide it. Sensitive health data also raises the stakes: a leak harms real people, and trust is hard to rebuild.
### The ontology
**Objects**
- 13 core things
- The nouns of the sector. Each object has an identity, properties and a history, and maps to records in the source systems.
- Patient
- Encounter
- Facility
- Clinician
- Diagnosis
- Prescription
- Medicine
- Stock batch
- Lab order
- Referral
- Appointment
- Program
- Equipment
**Links**
- 8 core relationships
- How the objects connect. Links turn separate tables into a graph that people and agents can walk.
- Patient has Encounter at Facility
- Encounter is recorded by Clinician
- Encounter produces Diagnosis, Lab order and Prescription
- Prescription dispenses Medicine from Stock batch
- Facility holds Stock batch and Equipment
- Referral moves Patient between Facilities
- Patient enrolls in Program
- Appointment schedules Patient with Clinician
**Actions**
- 6 governed actions
- What people and approved agents may do. Each action has rules, an approver where needed and a log entry.
- Raise a stock reorder
- Create and track a referral
- Schedule a follow-up and send a reminder
- Flag a patient for care review
- Redistribute stock between facilities
- Submit a facility report to the district
### Use cases
- Medicine stock
- A district health office manages essential medicines across dozens of clinics. Stock counts arrive monthly on paper or by SMS. Clinics run out of antimalarials or antibiotics while another facility holds surplus near expiry.
- Stock batches, consumption and expiry dates are modeled per facility. The system estimates when each clinic will run out and proposes transfers or reorders. Notices go to clinic staff by SMS in the local language.
- The district pharmacist approves transfers and orders. Clinic staff confirm what they actually hold, and the count corrects the model.
- Referrals
- A nurse at a rural health post needs to send a woman in obstructed labor to a hospital. She phones around to find a facility with a free bed, a surgeon and transport. The patient's file travels on paper.
- The referral object links the patient, the sending and receiving facility, and the clinical summary. The system shows which facilities report capacity, notifies the receiving team and tracks the outcome.
- Clinicians choose the destination and decide whether the case is an emergency. The system only displays capacity and keeps the record.
- Chronic care follow-up
- Patients on long term treatment for HIV, diabetes or hypertension miss visits. Clinics rely on a nurse to flip through registers and call people who have not returned.
- A list of patients overdue for a visit is built from encounters and appointments. A governed agent drafts reminders by SMS or WhatsApp and records replies. It flags patients who may have moved to another facility.
- The nurse in charge reviews the list, decides who needs a home visit and checks that messages respect patient consent and privacy preferences.
- Radiology and lab support
- Referral hospitals receive scans and lab results from many sources. Specialists are scarce, and reports back to remote clinics take days.
- Studies and results attach to the encounter. Approved decision-support models prioritize urgent studies and draft a structured summary, with the evidence and model version logged.
- The radiologist or physician reads the study and signs the report. A model output never replaces a clinical sign-off.
### Data you already have
- We read from your existing systems first and write back only through approved actions. Every source has a named owner. Where connectivity is weak or records are on paper, we plan for queued sync and OCR with human review.
- Electronic medical records and hospital information systems
- Paper registers and patient cards, digitized through OCR and data entry
- National health information systems such as DHIS2-style reporting
- Laboratory and radiology information systems
- Pharmacy, dispensing and drug logistics systems
- Program and registry data for HIV, TB, malaria and maternal health
- Equipment inventories and maintenance logs
- SMS, USSD and WhatsApp interactions with patients and community workers
### Compliance and residency
- Nigeria
- NDPA 2023
- Health data is typically treated as sensitive, which may require stronger consent and safeguards.
- Kenya
- Data Protection Act 2019
- Health data is sensitive personal data. Processing needs a lawful basis and security measures, and transfers abroad are restricted in some cases.
- South Africa
- POPIA
- Special personal information, including health, has stricter conditions. Sector health laws add confidentiality duties.
- Ghana
- Data Protection Act 2012
- Special personal data, including health, has additional conditions for processing.
- Rwanda
- Law on Personal Data Protection 2021
- Sensitive data rules and storage limits may influence where records and models are hosted.
- Health ministries, professional councils and facility licensing bodies may add rules on records and telemedicine. These examples are not legal advice. Confirm current requirements with counsel and your ministry. Examples only, not legal advice. Confirm with counsel.
### Boltcliq products
**Zeus**
- Model plane
- Models patients, encounters, facilities, medicines and referrals, and powers stock, capacity and program dashboards for district and national teams.
**Odin**
- Trust plane
- Gives clinicians, pharmacists and community workers access limited to their facility and role. Patient level access can be tied to purpose and recorded.
**Monolith**
- Control plane
- Limits what a health agent can read and do. A reminder agent can see appointments but not diagnoses, and a stock agent cannot touch patient notes.
**Argus**
- Runtime plane
- Hosts sensitive health data and agents privately, with the option to keep everything in country or inside the ministry's own environment.
**Cyclopes**
- Observation plane
- Logs who viewed or changed a record and which data fed a recommendation, which supports patient privacy reviews and clinical governance.
### Frequently asked questions
- Do you provide clinical decision software?
- No. We build the data and workflow layer. Any decision-support model is clinically owned, validated by your experts and always reviewed by a clinician.
- Can this work in facilities with poor connectivity?
- Offline-tolerant design is a stated requirement. Mobile and edge components should queue work and sync when a link returns. Exact support depends on the engagement.
- How do you protect patient privacy?
- Access is limited by role and purpose, sensitive fields can be masked from agents, and every access is logged. Hosting can stay in your jurisdiction.
- Will we need to replace our EMR or DHIS2 system?
- No. We connect to what you already use and add a model and workflow layer above it.
- What about paper registers?
- We plan for OCR and structured data entry, with people checking anything the system cannot read cleanly.