Importance of Artificial Intelligence & Biophotonic Techniques in Point of Care Diagnostocs of Hiv/Aids

Bio photonics and Artificial Intelligence Techniques

With the applications of optics and photonics, various methods are developed by spectrophotometer, microscopy, and lasers etc. for disease diagnosis in not only molecular level but also in tissue level. Some of them are described below:

• Hyper spectral Imaging (HSI): It is also called imaging spectrometer which involves various medical applications specifically in image guided surgery and diagnosis of diseases. It has been assumed that the scattering, fluorescence and absorption properties of tissues vary as the disease progresses. Therefore, the transmitted, reflected, and fluorescent light from tissue has been captured by HIS having quantitative diagnostic data. Which can be used by artificial intelligence based machine learning techniques to interpret HIV disease and use of ART in particular time [5].

• Diffuse optical Imaging: This technique is classified in two groups:

a) Diffuse optical topography—If the tissue optical properties get modified after a period of time, there are possibilities for the photon to reach same detector by altering the measurable intensity. A 2D topographical data set can be constructed by measuring the changes between every set of source and detector. This helps to study HIV at a tissue level [6].

b) Reconstructed topography—the absorption changes determined by high resolution images leads to production of 3D reconstruction of the image. It is usually done via the detection of the absorption distribution, where the data measured is matched with simulation results of numerous absorption distributions inside the 3D images. Hence, the recording of the signal at various distances between source and detector is essential and measured [7]. These results would be used input for Artificial Intelligence based machine learning techniques for disease diagnosis in particular stage as well as molecular level of HIV infection is well understood.

• Diffuse optical tomography: This techniques needs computed tomography in reconstructing the 3D images where recording of sequence of tissue measurements is performed. The 3D image has to be acquired from various angles which will give complete information of HIV affected cells/tissues [8].

• Lasers: Its better if Lasers system is used with sensor control. Because sensor controlled laser systems are of focus in the field of therapy. This was previously used in treating cancer. But it can be used for monitoring of HIV infection spread in body.

• Flow Cytometer (FC): A flow cytometer is a machine driven instrument that can be used to examine single cell properties i.e., allow only one cell to be analyzed at a time. It can be used to measure the cell granularity, cell size, and to quantify various cell components that include newly synthesized DNA and the total DNA, the number of specific cell surface receptors, gene expression as the amount of messenger RNA for a particular gene, and amounts of transient signaling events and intracellular protein in living cells. Hence FC is most adaptable to study HIV features. It is used in diagnosis of HIV disease as well as its progression [6].

• Fluorescent Markers: A significant feature of Bio photonics involves the visualization and detection of cells and tissues. Which includes injection of fluorescent markers, into a living system, to follow dynamics of a cell and drug delivery? This visualization and detection of fluorescent markers can be counted and effect on living system is shared by AI based ML techniques. By this method cell reaction and drug delivery to tissues are monitored in specific time. This can be used in Fluorescence lifetime imaging microscopy (FLIM). The information obtained by FILM is used in local environment sensing, detection of molecular interactions, detection of conformational changes, discrimination of multiple labels or background removal, tissue characterization by auto fluorescence and characterization and quality control of new materials. All these parameters based study are required for input data of machine learning techniques. And our expected outcome would able to predict which parameter is best for study cell based interactions of HIV. Confocal microscopy and multiphoton microscopy are also found to be suitable for HIV cells. Because in multiphoton microscopy, instantaneous absorption of two incident photons from a pulsed infrared laser source is observed. So this will helpful in how a particular HIV infected cell response [7].

• Targeted molecular imaging: It involves analyzing micron-level biological processes. It is used to analyze the shape and role of the molecular system by generating the signals incident from the molecules. Therefore the produced image describes the 3-D spatial distribution of the targeted molecules in the tissue, specifies the diagnostic data at the molecular level, and shows the functional cell properties. These properties are essential to study HIV infected cells at molecular level with its functionality opens new avenues of HIV [8].

Other methods like optical trapping, second harmonic trapping, and cell transfection are used to explore in case of HIV.

Artificial intelligence makes our computer to solve the complicated problem by training and testing the data given by us. Hence our intelligence and computer learning intelligence works together to develop the model or our knowledge for better use.

Machine learning which evolved from pattern recognition and computational learning theory, is able to construct algorithms that can learn and make predictions with data. There are many machines learning software tools. We use decision tree induction algorithms and Naive bayes algorithm of WEKA software package [6] to classify and compare a given HIV data set. CD4+count and IL-10,p24, IFN-biomarkers were used to determine diagnosis and screening of HIV/AIDS [9, 10]. There interaction and progression during HIV course was detected by above discussed bio photonics methods:

- Decision tree: A decision tree (or tree diagram) is a decision support tool that uses a tree-like graph or model of decisions and their possible consequences, including chance event outcomes, resource costs, and utility. Another use of decision trees is as a descriptive means for calculating conditional probabilities. In data mining and machine learning, a decision tree is a predictive model; that is, a mapping from observations about an item to conclusions about its target value. More descriptive names for such tree models are classification tree (discrete outcome) or regression tree (continuous outcome). In these tree structures, leaves represent classifications and branches represent conjunctions of features that lead to those classifications. The machine learning technique for inducing a decision tree from data is called decision tree learning, or (colloquially) decision trees [11, 12]. With the help of symptoms of HIV infection, and above mentioned biomarkers, a decision tree model is developed which would tell whether the person is infected with HIV and is in which stage. Accordingly antiretroviral therapy will started. As this is very fast and accurate it can be used in POC diagnostics for HIV/AIDS.

- Naive-bayes: This classifier is a simple probabilistic classifier based on applying Bayes’ theorem with strong (naive) independence assumptions. A more descriptive term for the underlying probability model would be "independent feature model". In simple terms, a naive Bayes classifier assumes that the presence (or lack of presence) of a particular feature of a class is unrelated to the presence (or lack of presence) of any other feature.

Depending on the precise nature of the probability model, naive Bayes classifiers can be trained very efficiently in a supervised learning setting. In many practical applications, parameter estimation for naive Bayes models uses the method of maximum likelihood [13, 14].

Other machine learning algorithms are also used based on the data provided by bio photonics techniques and accordingly model development and validation can be done for further investigation.