Midterm Flashcards
What aspects are distinct in biopharmaceuticals vs. drugs? (Biopharmaceutical)
- Large molecular masses (sometimes not even single macromolecules)
- Considerable (micro)heterogeneity
- Requires information on its biological source and processing
- Many levels of characterization
- Typically temperature sensitive
- Regulatory regimes unclear/inconsistent
- Systemic administration
- Patent: biological entities/processes
What aspects are distinct in biopharmaceuticals vs. drugs? (Drug)
- Relatively few atoms
- Consistent and standardized manufacturing
- Manufactured from chemical precursors
- QC procedures
- Shelf-stable
- Defined regulation
- Typically non-invasive administration
- Patent: molecular structure
What are the advantages of recombinant DNA technology over traditional methods?
- Overcomes problem of source availability
- Overcomes problems of product safety
- Provides alternative to direct extraction from inappropriate/dangerous sources
- Facilitates generation of engineered therapeutic proteins displaying some clinical advantages over native protein product
What are some of the hurdles to overcome for cell therapy manufacturing?
- Scaling up for adherent cell cultures
- Purification methods (magnetic bead separation contamination)
- Viral gene transfer
What are the drivers behind the rapid growth in AI health applications?
- Breakthrough in methods (deep neural networks) and causal interference approaches (structural causal models)
- Large datasets (EHRs)
- Diverse and multimodal datasets (genomic)
- Improvement in data standards (ICD-10)
- Improved data interoperability and healthcare data exchange
- Increased computing power
- Advancements in data privacy preserving approaches
A study on reproducibility showed that machine learning for health underperformed in all areas of replicability when compared to computer vision, natural language processing, and general machine learning: why is this the case?
Due to the sensitive nature of patient data, reproducibility can be limited for health applications of machine learning
Give some examples of where biases arise in AI for health applications and why is it particularly detrimental?
- Unfair sampling can skew model
- Human bias can interfere with data selection/weighting.
Detrimental in the case an AI application that could potentially affect millions of people with a biased evaluation.
In class, we discussed how adversarial examples are designed to cause a machine learning algorithm to make a mistake, what is the purpose behind this?
It is used to demonstrate the weaknesses in an algorithm, particularly in the context of ML applications that are intended to operate without human supervision (ex: EKG).
How can genomics, transcriptomics, proteomics, and metabolomics all contribute to the discovery of new therapeutics?
Genomics: characterization of inheritable disorders
Transcriptomics: measure fluctuations in levels of RNAs over time and in response to various perturbations
Metabolome: direct fxnl measure of proteome activity
Proteins cannot only vary their abundance but can also be modified - a process that can be specific to certain diseases or other influences, and may alter the functions of the protein. Therefore, elucidating such changes can lead to identification of key targets that, if modified, may have direct therapeutic effect or be indirectly indicative of a therapeutic effect.
Why is it important to have an independent test set for an ML model?
If an ML model is modified based upon the results of a test set, the modified algorithm cannot be modified with the same test set as it is no longer truly independent. This helps to ensure valid results.
Describe 3 contributing factors to aging.
- Stem cell exhaustion
- Telomere attrition
- Genomic instability
What is cellular senescence?
State of stable cell cycle arrest
Proliferating cells become resistant to growth-promoting stimuli
Typically in response to DNA damage.
What are some causes of cell senescence?
DNA damage
Telomere shortening
Dysregulated inflammation
Alteration of epigenetic modifications
Metabolic imbalance
Oncogene activation
Mitochondrial dysfunction
Production of ROS
Accumulation of excess metals
Describe how senescent cells can cause/contribute to disease.
Once senescent cell burden exceeds a threshold, self-amplifying paracrine and endocrine spread of senescence through the SASP outpaces clearance of senescent cells by the immune system.
Additionally, increased abundance of SASP factors may impede immune system function, further amplifying accumulation of senescent cells and subsequently accelerate other fundamental aging mechanisms, and contribute to age-related disease.
Describe the detrimental cellular senescent feedback loop.
Cellular senescence is induced following exposure to an initial stress. The resulting senescent cells produce a SASP that can potentiate further senescent cell accumulation and impair clearance by affecting immune cells. In turn, the SASP can be amplified, eliciting an environment of chronic inflammation and additional senescence-inducing stressors that drive a feed-forward cycle of senescent cell accumulation.
