Problem: Finding patterns in a table

Hello!
Thank you all for YOUR HELP in advance. Your assistance has been priceless. Unfortunately, I have encountered another problem that I'm unsure how to deal with.
Attached to this post is a file containing a small fragment of a measurement table showcasing changes in temperature for a specific physic-biological process under study. The complete table has dimensions of 600 columns and 1500 rows. For a better understanding of the problem, the table only includes differences in temperature changes, either as an increase by a certain number of degrees [C] or a decrease in temperature [C].

OVERALL DESCRIPTION:
? Rows indicate measurement points: from M1.1-M13.1 to M1.4-M13.4
? Columns represent measured values: S1-S50
? The numbers you are looking for are cell values marked in red.
? The desired changes in temperature values (red cells) are accompanied by:
1. Constant temperature increases, such as configurations of temperature increase by 1 degree (I marked these cells with a yellow color and bold black frame). The searched value corresponds to repeated constant configurations of temperature increase, for example: 0/1, 0/2, 0/3, 0/4, 0/5, etc.
2. Constant temperature drops, such as configurations of a temperature drop of 1 degree (I marked these cells with a bold black frame). The searched value corresponds to repeated constant temperature drop configurations, for example: 1/0, 2/0, 3/0, 4/0, 5/0, etc.
SAMP1.jpg
EXAMPLE:
Here's an example of the searched value: The temperature change between the measuring points M3.1 and M2.1 in column S14 corresponds to a temperature increase by 0/4 degrees [?C]. This increase is accompanied by a sharp temperature drop 7/0 (column S15) and a small temperature increase 0/1 (column S13) in the adjacent cells.

PROBLEM DESCRIPTION:
1. How to search the table and determine the potential set of searched cell values (cells marked in red or their location in the table) based on known patterns of changes in measured temperature values, for example: temperature increase with the configuration 0/1 or temperature decrease 1/0?
OR
2. How to automatically determine potential patterns based on already processed data? I have a part of the table where I have already determined the searched values using analytical calculations, rather than patterns of temperature increase or decrease. Analytical calculations are complicated, time-consuming, and I would like to avoid them.

I am asking for any idea how to solve the problem. If you have any questions, please contact me. THANK YOU VERY MUCH IN ADVANCE FOR YOUR INTEREST AND YOUR HELP!

Mark

I am unable to understand your pattern detection logic. It might help you think of programming strategies, but here's what I've done so far.

One programming strategy I use for this kind of thing is to reduce each possible pattern to a unique number or text string. Your example shows you taking a 2x3 block of cells and comparing the top row to the bottom row and determining if the top row values are larger, the same, or smaller than the bottom row values. There are 27 possible permutations, which I can represent as 27 unique numbers between 0 and 222 by using a formula like =111+100*SIGN(N5-N6)+10*SIGN(O5-O6)+SIGN(P5-P6) [using the references for your example, then copy/paste/fill into a parallel block of cells].

This scheme reduces your M2.1,S14 example to the number 220. I then scan the block of cells to see if you have consistently highlighted other 220 patterns, and it doesn't take me long to see a 220 pattern that you have also marked (M1.1,S15), and another 220 pattern that you did not mark (M11.1,S13). Obviously, I am not fully understanding the pattern detection logic you want to use, but perhaps it can still suggest a strategy you might use. Find a way to reduce each possible pattern in your detection scheme to a unique number or text string, then build a parallel block of cells that computes those values and you can analyze that parallel block.

After a few days, I will venture another response.

"Can it be done differently?" -- Short but useless answer, Yes, it can be done differently. I am unable to figure out the logic you are using, though, so I cannot offer any improvements.
"Can time shifts be included?" -- Again, yes, I'm sure they can be included, but I do not understand how you want to include time shifts so I cannot recommend anything.

If it helps, the strategy I proposed basically boils down to representing each possible pattern as a unique base 3 number. Your description of your examples is focused on trios of cells where each member (digit) of the trio can take one of 3 values -- "decrease", "no change", "increase." Thus, my proposal assigns each pattern a unique base 3 number from 0 to 222.

Some of your description suggests that size of change is part of this? Could we need base 5 numbers where each "digit" can take one of 5 values -- "large decrease," "small decrease," "no change," "small increase," "large increase" (resulting in base 5 numbers from 0 to 444)?

Basically, this strategy will require you to identify how many "digits" (cells, m) you want to consider in each pattern group, and how many possible "values" (conditions, n) each digit can have. Then your pattern detection algorithm allows you to reduce your raw data into a table of m-digit base-n numbers. Then search the "reduced" table for the m-digit base-n numbers that represent the patterns you are wanting to highlight. In this way, this strategy is very flexible, and you can implement a variety of conditions in your pattern detection. But it will require you as the programmer to clearly and carefully define what kinds of patterns you want to detect. At present, I do not understand what kinds of patterns you are trying to detect, so I cannot offer anything more specific.