Academics from The University of Sydney and Royal Melbourne Institute of Technology (RMIT) are part of a growing global field of food analysis that is finding new, effective ways to isolate, identify and quantify the beneficial components of health-promoting foods and natural remedies.
Today’s global food distribution system relies heavily on analytic tools to establish the safety, quality and potency of foodstuffs and nutritional supplements, enabling better regulatory enforcement and informing new product development. As consumers drive demand for ‘healthy’ foods (including so-called functional foods and nutriceuticals), these products are coming under increasing scrutiny. The need to find scientific backing for claims made regarding the safety, quality, potency and efficacy of so-called ‘health-promoting’ products, including traditional natural remedies, is fuelling a growing area of food analytics research.
A multidisciplinary group of researchers led by Dr Josiah Poon and Associate Professor Simon Poon, both from The University of Sydney’s School of Information Technologies, and Dr Daniel Sze from RMIT’s School of Medical Sciences, is at the forefront of this exciting global field.
Drawing on the vast clinical datasets of hospitals and treatment centres in Australia and mainland China, they’ve developed powerful data-analytic techniques for ‘fingerprinting’ several herbs commonly prescribed in traditional Chinese medicine (TCM), isolating and identifying the potency of their constituents and drawing meaningful connections that help determine their efficacy.
In collaboration with TCM and data-analytic experts around the world, the team has developed new, sophisticated data-crunching algorithms that address the specific challenges of this field – non-standardised diagnoses, highly personalised treatments, and a lack of tools with which to accurately analyse complex prescriptions – and enable them to generate reliable evidence-based predictions regarding the bioactive components of herbs commonly used in TCM formulae.
Dr Josiah Poon, who has been studying TCM for more than two decades, says that typically, eastern remedies are viewed as ‘non-scientific’. By using algorithms to analyse clinical records, his team has constructing the evidence base required to validate TCM’s scientific merit.
"We have been revealing the hidden knowledge buried deep within the massive clinical and textual datasets found in hospitals and/or healthcare clinics' records both here in Australia and China,” says Poon. “This kind of research can help confirm TCM diagnostic treatments as evidence-based, not just an ancient philosophy or subjective assessment.”
"We wanted to demonstrate how you can make use of machine learning, data mining, statistics and other analytic techniques to resolve research challenges,” says colleague Dr Simon Poon. “We also wanted to show how, if successfully applied, these techniques can provide insights on how to conduct future work on herbal or western medicines.”
"Researchers can [now] 'fingerprint' herbs, isolate and identify herb potency regardless of where in the world they are growing,” he explains. “We know herbs can vary from batch to batch or region to region, and consistency of potency has been an issue for those who question the legitimacy of TCMs. Previous analysis techniques have not been as extensive as those [now being employed].”
AgInnovators spoke with Dr Josiah Poon about the team’s research methodology, results to date, and the implications for those producing, consuming and prescribing TCM herbs and herbal remedies.
AG INNOVATORS: How are you analysing the herbs in your studies of TCM herbal remedies? What methods are you using to extract useful information from the databases?
DR JOSIAH POON: High-performance liquid chromatography (HPLC) is the technique employed in our study. The output from this technique is a chromatographic ‘fingerprint’. The fingerprint here is actually a graph, or chromatogram, of signal strength over retention time. Different herbs have different patterns, like our fingerprints; hence, we refer these patterns as the ‘fingerprints’ of different herbs.
The peaks in the patterns generally indicate the presence of compounds for an herb. HPLC is therefore a popular technique of quality control of herbal medicine.
AG INN: What are your most promising findings, especially with respect to bioactive ingredients in TCM remedies, and how do these findings advance work in the field?
JP: The fingerprint created by current HPLC methods represents the signal pattern of an herb over time in a particular frequency, i.e. a two-dimensional graph for a frequency. If a user chooses another frequency, a different fingerprint will be output.
Our technique differs by not only removing ‘noise’ from the input. Instead of coming up with a 2-D graph, a 3-D profile with richer content is created to represent an herb. This third dimension represents all frequencies that can be detected by the HPLC machine. We advocate this 3-D profile as a better fingerprint of an herb. The argument is that a user no longer needs to predetermine a specific frequency to find the peaks (compounds), when the interesting/useful compounds actually are ‘hiding’ in another frequency.
Preliminary work has also begun to associate the compounds in fingerprints of an herb with certain biological activities. The aim is to develop a method to predict the biological outcome, given a fingerprint. A journal publication is in preparation to report this result.
AG INN: How well are you succeeding in using science and technology to validate traditional insights and methods?
JP: A computational approach, on its own, is not sufficient to validate our result; hence, our team has been working with pharmacists/biochemists from the very beginning of the project. This enables our ‘discoveries’ in the digital lab to be validated by experiments in the wet lab. It is a crucial strategy to form a multidisciplinary team in order to make this research successful and useful.
AG INN: What's the next step?
JP: The work so far is to identify the relevant compounds – the peaks in a fingerprint – in a single herb, to predict certain biological activities. The result is interesting and it is useful for quality control. However, a TCM herbal treatment is not about a single herb but a set of herbs, boiled together, where new compounds may synthesise. The next step of our research is to investigate the therapeutic (biological) outcome owing to the synergistic effect of interacting herbs (effectively, the compounds).
AG INN: Is this a way for TCM and western medicine to ‘get on the same page’?
JP: Paraphrasing a saying from Deng Xiaoping: “It doesn’t matter whether a cat is white or black as long as it catches mice”, then: “It doesn't matter whether a treatment is TCM or western medicine as long as it heals people”. To understand the complementary effect of TCM and western medicine is an important step to advance integrative medicine. In fact, our team is working on a project with the Faculty of Medicine in the Chinese University of Hong Kong to study and understand stroke treatment using an integrative approach.
My personal goal is to investigate integrative medicine and to streamline the explanation from a system biology perspective.
AG INN: How is your research relevant to food producers, processors and distributors?
JP: The obvious answer is that we are all interested in the analysis of natural agricultural items. We are also interested in measuring the quality of and making predictions [about] certain outcomes of a given batch of items. In our case we use the output from HPLC, but it could as well be the result from any type of analysis.
Also, the prediction of efficacy in TCM is on a mixture of botanical products – which is very similar to, say, combining different fruits to yield a wine and predicting the aroma, colour and taste etc. In other words, TCM [herbs and] agricultural products can both be analysed using our MAGIC (Multi-dimensional Algorithms of Genomic Intelligence and Correlation) platform.
Further details of The University of Sydney researchers’ methodologies, challenges, findings and insights, along with those of worldwide experts in the field, can be found in an academic reference tool edited by Drs Josiah and Simon Poon and published internationally by Springer in February 2014.
Data Analytics for Traditional Chinese Medicine Research compiles the views of global experts in the field, who describe how data mining, machine learning and similar statistical techniques can be used to determine the potency and effectiveness of various herbs, and discuss how these have been or might be used to meet the challenges faced by researchers seeking to demonstrate the scientific credentials of TCM.
Various contributing experts in the field discuss how machine learning, data mining, statistics and other analytic techniques have been used to resolve research challenges, how successful or otherwise these techniques (if applied) proved to be, and how insights gained in the process might shape future work in the field.
The book will be of particular interest to those in the health informatics and data mining fields, biostatisticians, TCM and other health practitioners, and anyone involved in data analytics, particularly as applied to biologically beneficial components foodstuffs (including herbs).