The functional architecture design logic of the hospital information SaaS platform

Designing an efficient and user-friendly hospital information SaaS platform is not an easy task, it requires product managers to have an in-depth understanding of the medical scenario, starting from the real needs of each role from patients, doctors, nurses to administrative logistics staff, and building a system architecture that can not only meet the needs of hospital operations but also improve the patient medical experience. This article will delve into the functional architecture design logic of the hospital information SaaS platform, from electronic medical record management, medical resource scheduling, financial management to the optimization of the whole process of patient experience, and analyze how to achieve efficient and warm symbiosis of medical services through technical means.

In the deep waters of the digital transformation of the medical industry, the hospital information SaaS platform is no longer a simple electronic tool, but a nerve center that connects the entire chain of medical services. As product managers, we need to think outside the thinking of functional stacking, start from the essence of medical services, and design a system architecture that can not only meet the efficient operation needs of hospitals, but also allow patients to feel the temperature. This requires product managers to truly go deep into the medical scene – squatting in the outpatient clinic to observe the doctor’s information retrieval habits during consultations, following the nurse to record the pain points of repeated turnbacks in the nursing process, and even experience a complete medical treatment process from registration to medication collection as a patient. Only by thoroughly understanding the real needs of each character can the system cross the threshold from usability to ease of use.

This means not only to realize the technical linkage between modules, but also to break through the collaborative barriers between people, processes, and data. From the moment the patient steps into the hospital, the system should be like an invisible service butler: predict the need for examination during the outpatient clinic, synchronize nursing and doctor’s orders during hospitalization, and automatically split the medical insurance and self-paid parts when paying – so that the diagnosis and treatment process flows smoothly in the silent flow of data.

In the future, with the in-depth application of AI, Internet of Things and other technologies, the collaboration between modules will be smarter (such as predicting examination needs through patient historical data), but the core logic remains unchanged: let data run more, let people run less errands, and finally realize the symbiosis of efficient medical care and warm services.

1. Patient information module with electronic medical record as the core

The patient information module is the data cornerstone of the entire platform, but it is much more valuable than records. What we need to build is an information network that can grow dynamically with the patient’s entire diagnosis and treatment cycle, so that outpatient and inpatient data are no longer fragmented islands.

1. Outpatient files

The design of outpatient files should take into account the efficiency of initial filing and long-term reuse. For example, when a patient registers for the first time, the system can quickly obtain basic information such as name, gender, and date of birth through the ID card reader (supporting ISO14443TypeB standard), and automatically trigger a past history collection questionnaire (key items such as allergy history and surgical history are set as mandatory and are prompted by red stars). For elderly patients or children, family members are supported to temporarily obtain the permission to fill in through SMS verification codes, and the authorization time is 24 hours by default, and it will automatically expire when the time is over.

After 10 years of interaction design, why did I transfer to product manager?
After the real job transfer, I found that many jobs were still beyond my imagination. The work of a product manager is indeed more complicated. Theoretically, the work of a product manager includes all aspects of the product, from market research, user research, data analysis…

View details >

More importantly, the real-time linkage between outpatient files and the medical insurance system: when the patient presents the medical insurance electronic voucher, the system automatically verifies the insurance status (response time ≤ 300ms) through the API interface of the national medical insurance service platform, and synchronizes the medical insurance identity (such as medical insurance card number and insurance type) to all subsequent diagnosis and treatment links – when the doctor prescribes a prescription, the system automatically filters non-medical insurance drugs, and directly calculates the reimbursement ratio when paying to avoid the patient’s post-reimbursement process.

2. Hospitalization files

Hospitalization files need to focus on the granularity of procedural records. We once investigated in a tertiary hospital and found that nursing records in the traditional system often lead to incomplete information due to cumbersome entry. To do this, we designed condition-driven scenario-based templates:

• Postoperative patients: The system automatically generates timeline entries, such as blood pressure monitoring wound dressing change records every 4 hours, and associates the name of the surgery (such as after laparoscopic cholecystectomy);
• Obstetric mother: the template contains mother-infant linkage items such as postpartum blood loss and the number of neonatal lactations, and the nurse will synchronize the infant file when filling it out;
• Patients with chronic diseases (such as diabetes): Built-in blood glucose monitoring curve entry area, supporting manual input or automatic data upload via Bluetooth of the blood glucose meter.

Nurses only need to click on the time node to fill in the data, and the system will automatically associate it with the corresponding doctor’s order execution record (such as the execution time of cephalosporin antibiotic intravenous infusion compared with the temperature monitoring time) to avoid duplication of labor.

3. Integration of electronic medical records

The integration of electronic medical records is not a technical term, but a real improvement in diagnosis and treatment efficiency. We focused on solving three pain points:

• Automatic data aggregation: After the patient completes the routine blood examination, the system extracts key indicators such as white blood cell count and hemoglobin through natural language processing (NLP), automatically fills in the test summary area of the medical record, and uses linear regression algorithm to generate a trend chart for the past 3 months (such as a line chart from hemoglobin to 120g/L to 120g/L).
• Dynamic adaptation of permissions: When the emergency doctor receives a consultation, the system automatically opens the viewing permission of core medical records (including allergy history and underlying diseases) within 3 months based on the role, but it is prohibited to modify it. After the patient is transferred to the inpatient department, the attending physician automatically obtains the right to read and write complete medical records, and all modifications (such as changes in diagnostic conclusions) are generated through blockchain evidence storage to generate audit traces (including the ID of the modifier, timestamp, and hash value of the content before and after the change);
• Cross-hospital data exchange: By connecting to the regional medical cloud platform, the system uses the national secret algorithm SM4 to encrypt and access patients’ diagnosis and treatment records in other hospitals. Patients need to take the initiative to authorize after completing the face recognition authentication of the APP, and the scope of authorization can be accurate to the outpatient records or certain types of examination reports in the past 1 year, and the authorization records will be synchronized to the provincial health record platform for the record.

2. Medical resource module with supply and demand balance as the core

The scheduling efficiency of medical resources directly determines the service carrying capacity of the hospital. The core of the design of this module is to accurately match the needs of patients with limited resources through the combination of algorithms and rules.

1. Doctors who are both rigid and flexible are on duty

Behind the doctor’s scheduling system is a set of rigid rules + flexible adjustment. We implant multiple constraints in the underlying algorithm (based on genetic algorithm optimization):

• Chief physician: ≤ expert outpatient clinic 3 times a week, with an interval of ≥ 1 day (guaranteed hospitalization round time);
• Doctors in all departments: avoid continuous night shifts (intervals ≥ 12 hours) during the duty cycle, and the number of night shifts ≤ 8 times a month;
• Departments in short supply (such as pediatrics, emergency departments): 20% of the mobile outpatient quota is reserved, which is dynamically adjusted based on historical data (such as seasonal influenza periods).

More importantly, the emergency response mechanism: when the system detects a sudden increase in the number of registrations of a certain outpatient period by more than 30% (calculated based on 3 standard deviations of the data of the same period in history), it will automatically push additional consultation suggestions to the department director and simultaneously display the status of the deployable doctor (if Dr. Zhang is in the inpatient department, he can be seen after 1 hour Dr. Li is off today, whether it is an emergency recall). If the doctor confirms the additional consultation, the system will complete the whole process within 10 minutes:

1. the release number is sourced from the registration system (priority is assigned to emergency patients who have not been registered);
2. send SMS + APP push to the new source patient (including clinic number and estimated treatment time);
3. Adjust the consultation room (priority will be given to the allocation of vacant clinics on the same floor, and if the 3rd floor of the respiratory department is full, it will be transferred to the backup clinic on the 4th floor).

2. Equipment scheduling under high utilization

The difficulty of equipment scheduling lies in balancing utilization maximization with maintenance reliability. We create a full lifecycle profile for each device:

• CT machine: The average daily maximum load is set to 20 cases (including 1 hour buffer time), and if it exceeds it, the new appointment channel will be automatically closed (reminder to open the appointment at 8:00 tomorrow). This value comes from the compromise between the equipment manufacturer’s recommendation (maximum 25 cases per day) and the actual data of the hospital (lowest failure rate when 20 cases during peak hours);
• Nuclear magnetic resonance equipment: Due to environmental sensitivity, the system monitors surrounding vibrations (such as elevator operation, heavy equipment movement) through IoT sensors, and staggered the appointment time with the peak vibration (such as 9-11 a.m.) to reduce the image artifact rate.
• DR equipment: Establish a fault emergency chain, when a device reports an error (such as the tube temperature is too high), the system retrieves the spare equipment of the same model within 1 minute, calculates the transfer cost (such as equipment B is 50 meters away from the current clinic, 3 cases of remaining capacity), and pushes the option of changing the examination location + new time to the patient (support two time slots within 15 minutes/30 minutes).

Through this mechanism, the equipment utilization rate of a tertiary hospital has increased by 18%, and the waiting time for patient examination has been shortened to an average of 45 minutes.

3. Dynamic and intelligent drug inventory management

Drug inventory management should solve the contradiction between the risk of supply interruption and backlog waste. We use a dynamic safety stock model:

1) Commonly used drugs (such as amoxicillin): safety stock = average consumption in the past 7 days×3 (guaranteed turnover of 3 days). When the inventory is below the safety line, the system automatically generates a purchase order (for example, it is triggered when the average daily consumption of 50 boxes in the past 7 days and the inventory ≤ 100 boxes);

2) High-priced drugs (such as tumor-targeted drugs): adopt the zero inventory + pre-order model. When the doctor issues a prescription, the system queries the supplier’s real-time inventory (such as Shanghai Pharmaceutical’s osimertinib inventory) through EDI electronic data exchange, and generates a prescription after confirming that it is in stock, and the supplier goes directly to the pharmacy within 4 hours after the patient pays (the transportation trajectory is displayed in real time through GPS positioning);

3) Expiration Management: Built-in expiration date warning pyramid:

• 6 months before the expiration date: marked as concern (yellow label), priority is assigned to outpatient clinics (outpatient prescriptions are small, turnover is fast);
• 3 months before expiration: marked as an early warning (orange label), restricting new prescriptions, and only allowing inpatients to use it according to the current course of treatment (if the patient needs to take the drug continuously for 14 days, the system will automatically check the remaining validity period ≥ 14 days);
• 1 month before expiration: automatic freeze (red label), triggering the return process (generating a return order with an electronic signature, synchronized to the procurement system and supplier ERP).

3. Compliant and transparent financial management module

Hospital financial management is not only about settling accounts but also about data-driven resource optimization. The module is designed to be closely aligned with the principles of compliance and convenience.

1. Smooth registration charges

The key to the registration fee system is that there is no break point in the payment link. We have opened up online and offline omni-channel payment:

• Online payment: When the patient registers on the APP, the system automatically calculates the amount of medical insurance co-ordination + personal out-of-pocket payment, and supports the medical insurance electronic voucher to enjoy first and pay later (sesame credit score ≥ 650 points, monthly limit ≤ 3,000 yuan);
• Offline payment: The window is equipped with a QR code scanner + medical insurance card reader integrated equipment, and after the patient presents the WeChat/Alipay payment code, the system automatically splits the payment amount (such as the total cost of 200 yuan, medical insurance reimbursement of 100 yuan, WeChat payment of 100 yuan), and prints the bill with electronic signature within 3 seconds after the payment is completed.

For refund scenarios, a multi-level verification mechanism is designed:

• Patients who have taken the medicine: the pharmacy needs to scan the drug barcode to confirm that the drug has not been opened (packaging integrity verification);
• Patients who have undergone examination: the doctor needs to sign an electronic explanation that the examination has not been carried out (using CA certification signature);
• All refunds: Records are automatically synchronized to the financial audit system and reversed with the original payment path (WeChat Pay refunds are returned to WeChat change to avoid the risk of cash refunds).

2. Policy-adapted medical insurance settlement

The complexity of medical insurance settlement lies in the precise adaptation of policies. We have a built-in Drools rule engine in the system, which can automatically match the reimbursement ratio by region, type of insurance, and type of visit:

• Beijing Employee Medical Insurance Clinic: The part above the minimum payment line of 1,800 yuan will be reimbursed at 70%, and the system will display the cumulative amount of the minimum payment line in real time (such as the current cumulative amount of 1,500 yuan, and the remaining 300 yuan can enjoy the reimbursement);
• Medical treatment in different places: Realize policy penetration in the insured place through the interface of the national medical insurance service platform. For example, when a Shanghai patient seeks medical treatment in Guangzhou, the system automatically calls up Shanghai’s medical insurance catalogue (including drugs, diagnosis and treatment items) to determine whether ambroxol hydrochloride oral solution is within the scope of reimbursement (to avoid patients finding out that they cannot be reimbursed afterwards).

More importantly, the policy dynamic update mechanism: when the medical insurance catalog in a certain place is adjusted (such as adding/deleting drugs), the system will complete the rule update within 24 hours through the incremental synchronization interface, and retrospectively check the prescriptions in the past 3 months (if a drug originally belonged to Class A and was adjusted to Class B, the reimbursement amount will be recalculated, and more refunds will be made up).

2. Refined cost accounting

Cost accounting should be granular to the department and responsibility to the person. We break down hospital costs into fixed costs + variable costs:

1) Fixed cost allocation: The depreciation of large equipment (such as CT machines) is allocated to the department according to the duration of use, and the calculation formula is (original value of equipment – residual value) / estimated total hours of use × actual hours of use in the month;

2) Variable Cost Accounting:

• Labor cost: Calculate the doctor’s performance according to the length of the consultation (1 point per hour) + the number of operating tables (3 points for tertiary surgery);
• Consumables cost: bound to the prescription (such as a disposable infusion set used in a department, linked to the department code through a barcode).

The system generates a monthly cost-benefit analysis report for the department, which is visually displayed:

• Diagnosis and treatment income per square meter of consultation room area (such as 1200 yuan/㎡ for internal medicine clinics and 1800 yuan/㎡ for surgical clinics);
• the therapeutic effect brought by 100 yuan consumables (such as 100 yuan consumables in cardiology correspond to 3 effective treatment cases);
• The per capita diagnosis and treatment cost of outpatient clinics (such as 50 yuan/person for pediatrics and 120 yuan/person for orthopedics).

4. Administrative logistics module for refined management

Administrative logistics seems to be far away from the front line of diagnosis and treatment, but it is the invisible pillar of the stable operation of the hospital. The design of this module should focus on cost reduction and efficiency increase and risk prevention and control.

1. Controllable process and timely material procurement

We have built a three-tier procurement system that balances efficiency and compliance:

• Conventional materials (masks, syringes): Automatic replenishment, when the inventory is less than 7 days, the system sends purchase orders to fixed suppliers through electronic data interchange (EDI) (no manual approval) and synchronizes to the financial system to generate advance payment;
• Specialized materials (orthopedic implants): the department needs to submit an application, with a description of the necessity of use (such as the patient’s femur fracture needs titanium alloy steel plate), and after review by the equipment department (completed within 24 hours), the system will automatically associate the patient’s medical record number for easy traceability;
• Emergency materials (protective clothing): Trigger the green channel, the dean directly approves through the mobile terminal (supports fingerprint/face recognition), and the system synchronously notifies more than 3 suppliers to compare prices (price comparison time ≤ 30 minutes) to ensure that the goods arrive within 12 hours (docking with SF Pharmaceutical cold chain logistics).

To avoid the risk of rebates, the system automatically records supplier-purchase volume-price fluctuation data:

• When a supplier’s price is 5% higher than the average market price for 3 consecutive months (compared with data from platforms such as Yaozhi.com), a price warning is triggered;
• Dock with the hospital communication system, and if there is abnormal communication between the procurement personnel and the supplier (such as non-working hours calls ≥5 times/week), push early warning to the discipline inspection department.

2. Data-driven energy consumption management

Energy consumption management realizes the closed loop of data monitoring, abnormal early warning, and intelligent regulation. We deploy smart energy consumption terminals in all areas of the hospital:

• Outpatient hall: 1 light sensor is installed every 50 square meters, and the lighting system is automatically dimmed to 30% brightness when the natural light intensity ≥ 500lux during the day;
• Inpatient building: The air conditioner is controlled according to the needs of the department – the operating room has a constant temperature of 25°C (error ± 0.5°C), and the ward supports patients to fine-tune the ±2°C through the bedside screen (beyond the range requires nurse authorization);
• Equipment room: Current sensors are installed in the power distribution room to monitor the peak energy consumption of CT, MRI and other equipment in real time.

The system uses machine learning models to identify abnormal energy consumption points:

• The energy consumption of a ward suddenly increased by 30% at night: it may be that the air conditioner is not turned off, and the system pushes sound and light reminders to the nurse station (if it is not handled within 5 minutes, it will be upgraded to the head nurse);
• The monthly electricity consumption of a department increased by 20% year-on-year: generate an energy consumption analysis report, and judge whether there is equipment aging (such as a decrease in transformer efficiency) based on equipment usage records (such as DR machine start-up time increased from 8 hours to 12 hours).

Through these designs, the annual energy consumption cost of a provincial tertiary hospital was reduced by 15% (including 8% for intelligent lighting and 7% for air conditioning zone control), while reducing manual inspection workload by 20%.

5. The whole process of patient experience is seamlessly connected

When each module achieves deep collaboration, patients feel the non-probation service. We take registration-diagnosis and treatment-payment-medicine collection as an example to disassemble the system to make each link naturally connected.

1. Registration: intelligent diagnosis and path guidance

• Intelligent guidance: The patient enters a cough in the APP for 3 days, and the system uses the BERT model for intention recognition (accuracy rate of more than 90%), recommends the respiratory department, and displays the direction of the remaining 23 doctors today (such as Dr. Wang is good at children’s cough, Dr. Li is good at chronic cough in the elderly);
• Path guidance: After successful registration, the APP automatically generates a medical route map (the shortest path algorithm based on the hospital’s CAD drawings), marking that it is necessary to walk 200 meters from the entrance of the hospital to the 3rd floor of the respiratory department, and pass through the No. 2 elevator;
• Waiting reminder: The current number of people waiting for treatment is displayed in real time, and the expected waiting time is 40 minutes (calculated based on the average consultation time of the first 5 patients), and the upcoming number reminder will be pushed when the waiting time is more than half.

2. Diagnosis and treatment: information front-end and intelligent assistance

• When the patient enters the clinic, the doctor’s screen automatically pops up his electronic medical record through RFID positioning, and the right column is topped to display the previous allergy history of this complaint (such as penicillin allergy red);
• If the patient has unfinished tests (such as last week’s blood routine), the system will prompt whether to view the result first (support one-click jumping);
• When the doctor prescribes, the system checks the compatibility contraindications of drugs in real time (based on the database of the State Food and Drug Administration), such as cephalosporin + metronidazole combination will suggest that the risk of neurotoxicity may be increased; At the same time, it is recommended to replace medical insurance Class A drugs (such as the original imported ambroxol, domestic ambroxol is recommended, the price is 30% lower and 100% reimbursed).

3. Payment: Flexible splitting and multiple payments

• Segmented payment: After the patient completes the examination, the examination fee can be paid first (generate sub-order 1), and then pay the drug fee (sub-order 2) when picking up the medicine, and the system ensures that the item is not missed through order status management;
• Large installment: When the cost exceeds 5,000 yuan, the APP prompts the medical insurance reimbursement estimate of 4,000 yuan, and the self-payment of 1,000 yuan can be installed, docking with ICBC medical installment (0 interest rate for 3 installments, annualized 4% for 6 installments);
• Hybrid payment: Support medical insurance + self-pay combination payment (such as total cost of 1,000 yuan, medical insurance reimbursement of 600 yuan, WeChat payment of 400 yuan), and generate electronic bills (with the supervision seal of the Ministry of Finance) within 3 seconds after the payment is completed.

4. Medication collection: pre-preparation and peak staggering guidance

• Pre-preparation: After the doctor’s prescription is submitted, the pharmacy system automatically picks up the medicine through the intelligent sorting robot (sorting efficiency 300 boxes/hour), and calculates the estimated completion time (such as completing the dispensing at 15:30);
• Accurate notification: 10 minutes before the dispensing is completed, the APP pushes a reminder to the patient to pick up the medicine (including the pharmacy window number + medicine pickup code), and simultaneously displays the current window with 2 people queuing, and is expected to wait for 5 minutes;
• Peak staggering optimization: If there are 5 people queuing ≥ a window, it will automatically guide new patients to the adjacent free window (for example, 8 people queuing up in window 3, only 1 person is recommended in window 5).

Through this process, the average patient visit time was reduced from 120 minutes to 65 minutes, and the satisfaction rate increased to 92%.

6. Let each character have a convenient terminal

The work scenarios of different users vary greatly, and the terminal design must be tailor-made, rather than simply migrating.

1. Doctor Workstation (PC)

The core is the optimization of multi-use layout on one screen to efficiently process complex information:

• Left side (fixed area): basic information of the patient (name, bed number, admission diagnosis) and diagnosis and treatment timeline (sorted by examination-diagnosis-treatment);
• Middle (editing area): The medical record editor supports medical-specific voice input (using iFLYTEK medical model, with an accuracy rate of more than 95%, and supports professional term recognition such as atrial fibrillation);
• On the right (switchable panel): examination report (supports DICOM format image viewing), doctor’s order template (such as community-acquired pneumonia treatment plan), drug instruction manual (link to the State Food and Drug Administration database).

High-frequency operation setting shortcuts:

• F5 saves medical records, and Ctrl+R quickly recalls the last prescription;
• Ctrl+D calls the diagnostic template, and F12 prints the medical record (automatically attaches an electronic signature).

To reduce screen fatigue for doctors, it supports night mode (80% blue light filtering, brightness reduced to 30%) and automatic screen lock (encrypted after leaving the desk for 3 minutes, password/fingerprint unlock required).

2. Patient APP

The homepage only retains the three core entrances for registered report inquiry and payment (the font is enlarged to 16 for the elderly to view), and strives for a balance between simplicity and temperature:

• Report interpretation: The examination report not only shows the values, but also comes with popular interpretation (such as high white blood cells: there may be inflammation, it is recommended to follow the doctor’s advice), and support speech reading (Mandarin/dialect version);
• Remote assistance: Elderly patients can authorize their children (requiring face recognition) to help their parents register and view reports through the child assistance function (permission can be withdrawn at any time);
• Humanistic care: After the treatment, the system pushes rehabilitation tips (such as low-salt diet suggestions for hypertensive patients), and attaches the department phone number (working hours 9:00-17:00) for convenient consultation.

3. Nurse station terminal

Equipped with handheld PDA + ward bedside screen dual terminals to ensure mobility and fast response:

• Handheld PDA: The nurse scans the code (patient wristband + drug barcode) to confirm the execution of the doctor’s order, and the system automatically records the execution time + nurse work number, supporting offline operation (cache data when the signal is weak, and automatically synchronize after networking);
• Bedside screen: display the patient’s current nursing plan (such as temperature measurement at 10:00), click confirm and sign after the nurse completes (electronic signature is bound to the nurse’s practice certificate), and the data is synchronized to the electronic medical record in real time (using WebSocket real-time communication).

This design allows nurses to spend 30% less paperwork time (from 2 hours to 1.4 hours per day) and more energy to focus on patient needs.

The design of the hospital information SaaS platform is essentially a digital reshaping of the medical service process. It does not need disruptive innovation, but instills a patient-centric concept in every detail – let doctors write 1 minute less medical record, let patients run 1 less window, and let data take 1 more responsibility. When technology is truly integrated into the fabric of medical care, the symbiosis of efficiency and warmth can be realized.