How we helped a 7-café chain go from running blind to fully data-driven — read the real case study →
Home / Case Studies / The Machine Was Never the Problem
Healthcare2 staff, 50–60 reports/day, 7 test types

The Machine Was Never the Problem

See the SF LIMS product page →

Walk into a busy two-person pathology lab on a weekday morning and you will see the same thing everywhere in tier 2 and tier 3 cities: a queue of patients at the front, two staff members moving between sample collection and the testing machine, a printer warming up in the corner, and a PC with a folder full of Word documents that nobody outside this room has ever seen. This lab was doing between 50 and 60 reports a day. By any measure, that is a serious operation. The machine — a modern analyser capable of processing samples in one to two minutes — was doing exactly what it was supposed to do. The science was working. The problem was everything that happened after the machine finished.

PROBLEM

The machine result had to be retyped by hand

Every result the machine produced had to be manually copied — number by number — into a Word document. A Complete Blood Count has over a dozen parameters. A Lipid Profile has five. A Thyroid Profile has three values with specific decimal precision. Each of these was typed by a human, under time pressure, fifty to sixty times a day. The machine's output existed on its screen. The report's output existed on paper. Nothing connected them except a person sitting at a keyboard.

Twenty minutes of work for two reports

During a visit to this lab, the owner was observed completing two reports in twenty minutes. The tests themselves had taken under five minutes combined. The remaining fifteen minutes were spent locating the correct templates, clearing previous patient data, retyping results, and formatting the output for printing. At this pace, a lab doing sixty reports a day is spending roughly five to six hours on data entry alone — work that produces no clinical value and exists only because nothing connects the machine to the report.

No quality layer between machine and print

When a result is retyped by hand, the only quality check is the person doing the typing. There is no system comparison, no reference range flag, no second verification. A CBC result showing RBC at 13.7 where the machine said 13.0 goes straight to print. If the value is within normal range, the patient has no reason to question it. The referring doctor has no reason to question it. Nobody ever finds out. The error does not announce itself — it travels from fingertip to printed report with nothing in between. At fifty to sixty reports a day, the question is not whether errors happen. It is how many go undetected.

No audit trail, no patient history

When every report is a standalone Word document saved on a local PC, there is no connected record of anything. Looking up a patient who visited three months ago means searching through a folder of files named in whatever convention felt logical at the time. Checking whether a result was corrected after printing means finding a physical copy. Understanding trends across a patient's repeat visits — a rising TSH over six months, a changing lipid profile — is effectively impossible. The data exists in fragments. None of it talks to any other part.

Everything lived on one local machine

The entire operational history of this lab — every report, every template, every patient record — lived on a single PC with no backup system. A hard drive failure, a power surge, a corrupted Windows update: any of these ends the record permanently. There is no cloud copy, no version history, no recovery path. The lab had been operating for years. All of it was one hardware failure away from disappearing.

WE BUILT

The machine was never the bottleneck. The seven steps between the machine result and the printed report were. SF LIMS replaces every one of them — except the step that actually requires a pathologist. The workflow comparison below shows exactly what changed and where.

Before SF LIMS
After SF LIMS
01
Patient arrives
Name + tests written on paper slip
01
Patient arrives
Registered digitally in SF LIMS
02
Vial labelled by hand
Name, age, test — all handwritten
02
Test order created
System generates unique sample ID
03
Sample goes to machine
Machine runs in 1–2 minutes
03
Sample goes to machine
Machine runs in 1–2 minutes
04
Find the right Word template
One of 20+ files on a local PC
04
Sample ID fed into SF LIMS
Result pulled in automatically
05
Delete old patient data
Previous report still inside the file
05
Reference ranges checked
System flags values outside normal
06
Retype result by hand
Number by number, from screen to keyboard
06
Pathologist verifies + releases
Clinical judgment — not data entry
07
Format and print
~10 minutes per report. No record kept.
07
Report sent via WhatsApp
Auto-generated. Audit trail logged.
Machine Integration — The Core USP
Steps 4–7 are where SF LIMS changes everything. The machine result enters the system via sample ID — no human retyping, no Word document, no quality gap. The pathologist verifies. The system does everything else.
RESULTS

Machine output goes directly into the report

The result the machine produces is entered into SF LIMS via sample ID — no retyping, no copy-paste, no Word document. The transfer from machine to report is now a system function, not a human task. The gap that caused every transcription error is closed.

From 10 minutes per report to under 2

With patient registration, test ordering, and report generation handled by the system, the time between a machine result and a delivered report has collapsed. The pathologist's job is now clinical verification — not data entry.

A quality layer that did not exist before

Every result now passes through SF LIMS reference range checks before release. Values outside normal range are flagged automatically against age and gender banded norms. The pathologist reviews on screen and explicitly releases — nothing goes to print silently.

Full audit trail on every report

Every report now has a complete record: who registered the patient, when the sample was collected, what the machine returned, when the report was released, and when it was delivered. Corrections are logged with a mandatory reason. Nothing is overwritten silently.

Patient history accessible across visits

Every report is stored in the cloud and linked to the patient's permanent record. A returning patient's previous CBC, LFT, or Thyroid Profile is one search away — not buried in a folder of Word documents on a local PC.

Reports delivered via WhatsApp — no printing required

Once the pathologist releases a report, it is sent directly to the patient's phone via WhatsApp. The printer still works for those who need a physical copy. But for most patients, the report arrives before they have left the building.

SUPPORTING DATA
Pathology Lab Digitisation — Industry Context
Small labs still using manual or paper-based reporting70%
Diagnostic errors attributable to transcription and manual entry23%
Estimated figures based on industry reporting on diagnostic lab digitisation in tier 2 and tier 3 cities in India. The transcription error figure reflects published research on manual data entry in clinical settings. Neither figure is a measurement of this specific laboratory's performance.
Your project could be next

Tell us what you're building.

We take on a limited number of engagements at a time so every one gets senior attention from start to finish.