ComBase broth models can be used to make predictions in either a static temperature or under fluctuating (changing) temperature conditions. They can provide up to four simultaneous predictions and include growth and survival curves as well as thermal and non-thermal death curves

1.- GENERATING A SINGLE PREDICTION AT STATIC TEMPERATURE

Select the model category: select 'growth model' for prediction of bacterial growth or 'thermal death model' for prediction of thermal inactivation.

Select the required model: The first step in producing a prediction is to select a model. A drop-down menu of models can be viewed in the ‘select a model’ listbox. Click on the arrow to view the full list. The model required for the organism of interest can then be highlighted and selected from the menu. Many of the models encompass three environmental factors (temperature, pH and water activity). Some models also have an additional fourth environmental factor e.g. CO₂ or lactic acid. One basic principle of empirical modelling is that one should not extrapolate: predictions should not be made outside the region of observations. The limits of the selected model for each environmental factor appear below the input fields.

Input environmental factor values: Default values for the environmental factors (temperature, pH, water activity, etc) automatically appear in each of the input fields. Default values should be replaced by values of interest to you. The range within which input values must fall is indicated beside the input field for each of the factors. If values outside the range are selected, an error message will be generated. Water activity can be expressed in terms of sodium chloride (%NaCl) or water activity. Water activity can be calculated from NaCl concentration by toggling between NaCl and Aw buttons (it is assumed that the salt is dissolved in water). The %NaCl values are transformed into water activity values by the formula:

Aw=1-%NaCl*(5.2471+0.12206*%NaCl)/1000 (Resnik and Chirife, 1988)

To get a prediction:When a four-factor model has been selected, values are also required for the 'Factor 4' field. The default value for the initial count is log₁₀cell concentration = 3 (i.e. 10³ cells/ml). For the physiological state, the default value is the value which was typical for the curves providing the base for the models. It is not possible to select ‘initial level’ values for thermal death models as results are provided as a relative decrease in log concentration.

The prediction output: The right-hand output panel now shows a graphical representation of the prediction. A statement of maximum growth rate and doubling time (or maximum death rate and D-value) are presented below the graph. Additionally, each of the time vs. cell concentration data points for the prediction are listed below the graph and may be selected and copied for use in other applications (e.g. Excel).

2.- GENERATING MULTIPLE PREDICTIONS AT STATIC TEMPERATURE

Multiple predictions are generated in a similar manner to single predictions. Click the 'add prediction' button up to three times to create multiple input sections. A total of four simultaneous predictions can be made. Using this feature, it is possible to compare the responses of a number of different organisms to a single set of environmental conditions or for a single microorganism to varied environmental conditions. To return to a single prediction, click the 'remove last row' button the required number of times.

3.- GENERATING A SINGLE PREDICTION UNDER FLUCTUATING TEMPERATURE

The tool does not only allows prediction to be made for static temperature conditions but also predictions under certain fluctuating (changing) temperature conditions. For example, predictions could be made from data representing the fluctuating time vs. temperature profile expected or measured for the following processes in sequence: the later stages of cooling, storage within the factory, storage during transportation, storage at retail premises, possible purchaser temperature abuse followed by domestic storage for the organisms Listeria monocytogenes and non-proteolytic Clostridium botulinum using the appropriate growth models. This 'fluctuating temperature' feature may also be useful to demonstrate the affect of a thermal process using logged data which falls within the relevant temperature range of the thermal death model.

Select Dynamic prediction:Input the temperature profile in the Time/Temperature table. For details about how to input your temperature profile, refer to the Data Input section

Select the required model: Select the required model as described for static temperature predictions.

Input environmental factor values: Values for environmental factors for ‘changing temperature’ predictions should be entered as described for ‘static temperature’ predictions except for the values pertaining to temperature.

The prediction output: The right-hand output panel now shows time vs. cell concentration data for the given prediction (white) alongside the selected time vs. temperature profile (green). No statements of maximum growth rate or doubling time are provided as these fluctuate according to the time vs. temperature input. However, each of the time v. cell concentration data points for the prediction are listed on the "Data Points" tab and may be selected and copied for use in other applications.

GENERATING MULTIPLE PREDICTIONS AT FLUCTUATING TEMPERATURE

This process is performed in a similar manner to that already described. This facility can be used to compare organisms or key parameters for a single fluctuating temperature profile. It is not possible to make predictions when the temperature ranges of the models do not overlap.

REFERENCES

Resnik, S. L. and Chirife, J. (1988). Proposed theoretical aw values at various temperatures for selected solutions to be used as reference sources in the range of microbial growth. Journal of Food Protection 51, 419-423.

For further information, read the user manual

1. What is the Phys. state value?

The Phys. state (or “initial physiological state”) is a dimensionless number between 0 and 1 expressing the physical suitability of the cells to their environment. If its value is 0, then growth will not occur (infinite lag); if the value is 1, then growth will commence immediately, without lag.

The duration of the lag depends not only on the actual environment (temperature, pH water activity, etc) but also on the history of the cells. Stresses (e.g. thermal, osmotic, acid) can significantly increase the lag times. An example of this history dependence of the lag duration is given in the chart below. Listeria monocytogenes was grown in the same medium at 15°C after sub-cultures from different conditions. The growth rate is similar in both cases but significant differences in lag durations are observed.

The phys. state value quantifies the effect of the history on the bacterial lag time. The lag time can be derived from the phys. state value using the formula:

lag = -log(Phys. state)/Max.rate

Figure 1. Growth of Listeria monocytogenes in broth, at 15°C after different subculturing procedures.

2. Which value should I input for the Phys. state value?

The default physiological state is set to 1, meaning already adapted cells, i.e. no lag time. Because the user is rarely able to provide its value, the user can set it to a value typical for the experiments used to develop the model. The user might experimentally evaluate the phys. state value suitable for the pre-incubation conditions of interest. This can be done by fitting growth curves obtained after incubatory conditions simulating the history of the cells (e.g. stress). Once the lag and the growth rate are calculated, the Phys. state can be deduced by the equation:

Phys. state= 10^(-lag x Max.rate)

It is suggested that users try different values for the physiological state, to study its effect on the growth curve.

1. Format requirements:

The time/ temperature should be inserted in the texbox on the 'changing temperature' page in the format illustrated and described below:

For each time temperature record of your profile, time is to be entered first (in hours) then temperature (in °C).

You must use the 'point' symbol "." as the decimal separator.

The profile can be typed directly or may be copied from another application (e.g. Excel spreadsheet, text file) and pasted.

2. Modelling requirements:

All temperature values must be within the range for the models selected.

There must be a minimum of 4 and a maximum of 100 (time vs. temperature) records in your profile.

The records must be recorded in chronological order.

The first time-point must be '0'.

The last time point must be less than or equal to '5000' hours.

Revised Models

Until recently, the first version of the ComBase Predictor had a problem with the broth four-factor growth models: under certain combinations of conditions (like high CO₂ or added preservatives), it might have shown bacteria growing faster than under less stringent environmental conditions —when in fact, they shouldn't. This happened only under conditions around which there were no measured data (extrapolations). Thanks to feedback from users, these problems were identified.

The ComBase Predictor update (v1.1), corrects this by applying a stricter mathematical “clean-up”, called backward elimination, of non-significant terms in the formulae. This ensures that predictions now follow a more biologically sound and expected (from experience) pattern.

While the first version, ComBase Predictor v1.0 served its purpose well, ongoing research and user feedback have pointed to directions for further developments. This is natural in predictive sciences, much like in weather forecasting: models evolve as we learn more and refine the methods. The updated version, ComBase Predictor v1.1, brings a more reliable mathematical treatment of the data, especially when multiple environmental factors are involved. Users can expect further improvements within a year or two as development and refinement continues.

Summary of Revisions

For complete details, refer to the following PDF: ComBase Predictor Update December 2025.