Healthcare · Data platform + forecasting

Forecasting seasonal patient surges, so beds and staff are ready in time.

A hospital moved from reacting to seasonal disease surges to planning for them, using a forecast built on its own admissions history.

Before

Each department kept its own records in separate systems and spreadsheets. Leadership couldn't compare departments side by side. Seasonal disease outbreaks were handled after the fact: beds and staff were added only once wards were already full.

What we built

A lakehouse that brings every department's patient and operational data together. Records were cleaned, standardized and de-identified, then organized into one view of admissions by department, diagnosis and month. On top of that: department dashboards and a forecasting model trained on several years of admissions history.

What we found

Seasonal diseases followed a clear, repeating pattern, with cases rising sharply from midsummer through the rainy season. Once the data lined up, the pattern was impossible to miss.

After

  • Management plans for seasonal surges instead of reacting to them.
  • The forecast shows how many extra beds and health workers each affected unit will need in the months ahead.
  • Every department is compared on one dashboard.

Stack

  • Databricks
  • Python
  • Multiple linear regression
  • Power BI

Illustrative

Monthly seasonal admissions over several years, with the forecast extending ahead and a shaded beds needed band beneath it.Illustrative, with demo data. A line of monthly admissions over past years rises and falls in the same season each year. A dashed forecast line continues the pattern into the months ahead, with a shaded band beneath it showing the beds needed.Monthly admissionsPast yearsForecastBeds needed
Admissions rise in the same season every year, so the forecast shows the beds needed before the surge.

More case studies

Ready to see where your data stands?