entropy with previous biological data slide 03-24-2014d1520 slide -"C:\Users\kurtw_000\Documents\kurt\storage\CIM Research Folder\DR\2014\03-24-2014d0913\entropy with previous biological data 2.pptx" -"C:\Users\kurtw_000\Documents\kurt\storage\CIM Research Folder\DR\2014\03-24-2014d0913\entropy with previous biological data.pptx" some powerpoints this slide is found in C:\Users\kurtw_000\Documents\kurt\storage\CIM Research Folder\kwhittem\Presentations\2014\PhD Oral Defense\Oral Defense Presentation Kurt Whittemore 4-11-14.pptx some text from slide { Summary points -Cancer (cancer symbol is zodiac symbol for crab) --DNA copy number, and mutations --alternative splicing -chromosomal abberations in karyotypes --protein interaction network --photographs of tumor tissue --thermal images of body parts -Heart --heart rate complexity decreases with age while watching Fantasia Disney movie (more pronounced in males than females) --ECG data during sleep has lower entropy with higher age --magnetocardiography data entropy was lower in patients with coronary artery disease -Brain --increase in fMRI data entropy with age (males higher than females) --schizophrenic patients had lower entropy than normal people --increase in MRI image data entropy with age and alzheimer's --increase in linguistic processing entropy --more entropy in older individual's ability to remember the location of an object --increased neuronal firing entropy in Parkinson's. High frequency stimulation decreased entropy -Immune system --the entropy of virus types present in blood decreased after immune suppression after an organ transplant -Age --DNA methylation --many systems increase in entropy with age -Trends --heart is exception --entropy increases with age --high entropy indicates a disease state --entropy that is too low indicates a disease state (more rare than high entropy) -Entropy in certain range correlated with health states, and entropy shifts away from this range with age Some text from dissertation associated with this content { Using the concept of entropy to provide quantitative correlations with health and disease states has also been explored previously by many other researchers. The concept has been used with data associated with cancer, the heart, and the brain. The entropy of this various biological data has also been investigated with regard to age, and changes during the aging process have been observed. These previous research projects provide a foundation and framework of concepts which the AbStat can integrate into. Ultimately, the entropy measurement can be used to assess the health of many different biological systems, and the immune system is an ideal candidate system for health assessment. There are several papers relating the entropy of biological data with cancer. Many datasets have been analyzed, and the result demonstrates that genomic entropy calculated from aberrations in DNA copy number is higher in a variety of cancer types 25. Additionally, the increase in genomic entropy is correlated with an increase in gene expression entropy. A separate mathematical study claimed that the increased mortality rates and cancer rates with age observed in models such as the Gompertz equation, Weibull function, and Strehler-Mildvan modification of the Gompertz equation can be associated with increased informational entropy of the genome with time 28. The author also suggests that increased genomic informational entropy can even be caused by mutations induced by thermal noise over time in the absence of chemical mutagens and radiation. In addition to DNA focused studies, the entropy of RNA has also been investigated. The entropy of alternative splicing in cancer cells was explored, and the results demonstrate that the alternative splicing entropy is significantly higher in 13 of 27 cancers investigated when compared to normal tissues of the same anatomical site 24. The distribution of splicing isoforms present in cancer cells was much broader and flatter, resulting in a higher entropy. Interestingly, the genes that presented the highest entropy in their splicing isoform distribution are splicing factor genes themselves. The study also found that there was a positive linear correlation (correlation coefficient of 0.81) of proliferation level and entropy value for tumors. If we zoom out from the DNA and RNA level, we find that the entropy of the cell at higher levels has also been determined. In one study, the entropy of structural and numerical chromosomal aberrations were compared among 14 solid tumor types in 1,232 karyotypes 26. Some cancer types such as lung cancer typically had higher entropy values than other tumor types, and high entropy values were associated with a shorter mean survival time for the patients. In another study, researchers discovered that the entropy of a random walk on the protein interaction network graph was higher in cancer cells than normal cells in all of the 6 different cancer tissue types investigated 29. Cancer data at the tissue level has also been examined. Histological analysis of photographs of tumor tissue at various magnifications also benefits from the use of entropy as a measure to distinguish normal from tumor tissue 27. In the histological study, three measures were suggested as particularly useful for diagnosing a sample as a tumor sample using tissues from 34 prostate tumor samples, 34 benign hyperplastic samples, and 34 normal prostate samples: fractal dimension, cell nuclei number, and entropy. All three of these measures were determined from an image file of tumor tissue. Researchers have also calculated the entropy of data from thermal images of women with or without breast cancer. They were able to use these results and other attributes of the thermal image along with classification algorithms to classify samples as cancer better than would be expected by chance 30. Researchers have correlated health conditions with the entropy of biological data related to the heart. In one study, researchers discovered that the “band limited transfer entropy” decreases with age as heart rate complexity also decreases 31. In the study, the heart rate, respiration rate, and blood pressure were monitored for 20 young subjects aged 21-34 years old and 20 older subjects aged 68-85 years old as they watched the movie Fantasia from Disney. The researchers then used sophisticated non-linear dynamics techniques to calculate the joint entropy and conditional entropy to determine the contribution of respiration and blood pressure data to correlate with heart rate complexity. They found that the lag between respiration and heart rate was longer in older subjects and the entropy calculated from the data was decreased in older subjects. This age associated effect was more pronounced in the males than the females. Another group found that the spectral entropy of electrocardiogram (ECG) data recorded during sleep was negatively correlated with age 32. In another study, the researchers calculated the entropy of magnetocardiography data, input these entropy values into a multilayer perceptron neural network as training data, and then classified whether heartbeats were from patients with coronary artery disease or normal individuals with 98% accuracy 33. The entropy of one of the regions of interest in the magnetocardiography data was considerably lower in the patients diagnosed with coronary artery disease. Several researchers have used entropy associated with biological data for the brain. One group correlated an increase in entropy in fMRI data with age in a very large dataset with 1,248 samples 34. The entropy measured the dispersion of the functional connectivities that exist within the brain. In addition to a correlation with age, they found that males exhibit a higher increase of entropy with age then females. They also discovered that schizophrenic patients had a lower functional entropy than normal people. This result illustrates that normal individuals fall within an entropy range, and a value too high or low from this range can indicate a deviation from optimal health, and this also turns out to be the case with AbStat as well. Another MRI study investigated the entropy of the cortical structure complexity determined from images constructed from MRI brain data 35. They determined that this entropy increases with age, and the value of the entropy was also higher for Alzheimer’s patients compared to age-matched controls. In a different study, researchers measured the ability of the brain to process linguistic information at different ages 36. The researchers asked old and young adults to identify and repeat words which had high or low response entropy. For example, a word with a very high response entropy would occur in a neutral context such as in the sentence: “The word is __.”. The final word in this sentence could be many different words resulting in a very large degree of uncertainty. On the other hand, a word with a very low response entropy would occur in a sentence such as “He wondered if the storm had done much __”. Based on the responses from 100 previous volunteers, the final word in this sentence has a very high probability of being “damage” so there is less uncertainty (and lower response entropy) in this “high context” situation. Older individuals failed to identify the last word of the sentence more often than younger individuals even in low entropy response situations, and the researchers make the argument that this is not due to the loss of hearing acuity that occurs with age. In short, the concept of entropy can also be correlated with the ability of the aged brain to process linguistic information. In a more visually oriented study, older individual’s display a higher variability in an attempt to remember the location of an object, and the researchers claim this is due to a higher entropy in neuronal processing 37. They create some new terms for their molar entropy model for spatial memory and suggest that the slower processing speed and less accurate outcomes of older individuals is caused by increased neuronal noise or “computational temperature”. In one study, the researchers did more than just monitor the entropy of the brain, they intervened and changed the entropy. In the study, they found that rhesus monkeys with induced Parkinson’s disease exhibited an increased level of neuronal firing entropy in the subthalamic nucleus area of the brain compared to the entropy in the control group of rhesus monkeys 38. They then implanted an electrode into the brain and induced high frequency stimulation or low frequency stimulation. High frequency stimulation resulted in lower entropy and a reduction of Parkinson’s disease symptoms, whereas low frequency stimulation increased entropy and exacerbated symptoms. In addition to some of the studies mentioned previously which associated entropy measures with age, there are also other measures. One biological characteristic that increases with age is the entropy of DNA methylation. Researchers demonstrated that the methylation changes of the DNMT1 gene, which itself codes for a DNA methyltransferase, exhibited aging-driven entropy characteristics in a White Leghorn chicken model 39. Another author has discussed the general concept of an increase in entropy among many biological systems with age 23. There are a few trends in these research articles which address the entropy of biological data. In all of the studies, the value of entropy increases with age, with the exception of the heart studies. The complexity of heart rate data actually decreases with age. Additionally, higher than normal entropy values are correlated with disease states such as cancer. Normal healthy states are demonstrated or implied to fall within a certain entropy range, and values that are too far above or below this range indicate a problem with the system. As an example of an entropy value that was too low, schizophrenic patients exhibited lower than normal entropy values associated with fMRI data as mentioned previously 34. All of this collective information supports the ideas that entropy is an important measure correlated with health states, healthy systems exhibit entropy within a certain range, and that entropy tends to shift away from this normal range with age. The entropy measure has been applied to many different types of biological data in numerous studies, but the AbStat measure is unique for a few reasons. The first reason the AbStat measure is unique is that it combines entropy with many other global measures to boost the power of classification and diagnosis even further. This type of combination with this many global attributes was not performed in any of the articles previously reviewed. The second reason that the AbStat is unique is that it determines the entropy calculated from the interactions of antibodies with a large number of peptides rather than from the original source of the health problem. The nature of the response of the immune system can often determine the outcome of the patient. Since the immune system plays a very large role in the health of the patient, and since the immune system responds to many different problems in the body, using the immune system to provide information about the body is a logical strategy. } Human heart image http://www.flickr.com/photos/patrlynch/450128330/ Human brain image http://www.flickr.com/photos/11304375@N07/2041938587/sizes/o/ Cancer image http://en.wikipedia.org/wiki/File:Cancer.svg Cell Image http://www.flickr.com/photos/90500915@N05/9524577194/sizes/l/ Age Image http://www.flickr.com/photos/dinoowww/4557829098/sizes/m/ }